probe(review): fp8 serve row + 3-ref fixture + fp8 quality-collapse diagnosis

Review follow-up on PR #4: run --quantization fp8 through the served
path on the 3090 (the flag issue 1 validated for capacity), add the
untested 3-reference fixture, and measure all counts.

Capacity: confirmed — all of 1/2/3/4 references fit and run (4-ref
44.1-44.4s, server peak 23,522MB, nvidia-smi max 23,983MiB; Marlin
W8A16 is faster than BF16, 3.3x the Diffusers nocache baseline).

Quality: every fp8 output is structurally broken on this commit and
card (~97% transparent RGBA, grey RGB, PSNR 7-9dB) — on the edit path
at every reference count and on T2I alike (issue 1's fp8 artifacts
reinspected: same collapse; that run validated rc+image, not output
structure). img_mlp exclusion does not fix it; transformer-only fp8
also collapses, pinning the culprit on the Marlin-quantized DiT.
3090 verdict stays not-built; upstream fix order is fp8 output
correctness before any capacity role for it.

.agents/skills/3d-animation-short-generator/SKILL.md

@@ -0,0 +1,64 @@
+---
+name: 3d-animation-short-generator
+description: Develop stylized 3D narrative shorts through story, character and environment references, timed shot tables, storyboards, generation, and assembly. Use for coherent animated storytelling or its planning stages, not isolated image edits or photorealistic live action.
+metadata:
+  compatibility: File-based agent workflow with sibling h3-prompt-writing. Media generation and editing require available tools or a configured backend.
+---
+
+# 3D Animation Short Generator
+
+Read the [portable workflow](../h3-prompt-writing/references/portable-workflow.md).
+For a complete film, progress from story to references, shot table, text storyboard,
+clips, and assembly. Keep these as local artifacts, using existing user decisions
+and authorization. If only a planning stage is requested, deliver that stage.
+
+## Story and visual contract
+
+Capture premise, desired emotional outcome, duration, ratio, model, and audio mode.
+Use warm stylized 3D as a default only when style is unspecified: readable geometric
+silhouettes, designed hair clumps, tactile materials, soft skin shading where
+appropriate, expressive brows/eyes, and elastic posing. User-selected 2D or other
+styles override this preset rather than receiving accidental 3D texture.
+
+Create an active protagonist with a want, obstacle, meaningful choice, and payoff.
+Connect the ending to an earlier visual/emotional anchor. Use as many beats as the
+duration needs; do not pad a tiny story into an eight-beat formula. Dialogue should
+change a relationship or move the action, and use the user's requested language.
+
+## Reference design
+
+- Character references lock name, silhouette, proportions, face, hair, costume,
+  signature props, and color. Label identity in the production document; clean
+  rendering anchors should not carry labels that could leak into the video.
+- Scene references contain environments only unless the user requested a composite.
+  Record landmarks, lighting direction/time of day, entrances, exits, and prop state.
+- A reference board with multiple views is for design review. Prefer appropriate
+  clean views for model input so grids and pose sheets do not become final frames.
+- When an asset changes, update all affected prompts and clips, with a manifest
+  identifying the current revision.
+
+## Shot planning
+
+Read [shot-table-spec.md](references/shot-table-spec.md) for the six-column table
+and continuity checks. Then read
+[storyboard-guidelines.md](references/storyboard-guidelines.md) for the matching
+text storyboard and optional pencil previews. The text storyboard is the source
+of shot timing; preview art does not silently replace it.
+
+Vary shot size and elastic performance to serve the story. Camera tilts, close-ups,
+or exaggeration are choices for specific beats, not quotas. Track off-screen
+characters and spatial landmarks so they do not teleport back into later shots.
+
+## Generation and assembly
+
+Before rendering, read [model-selection.md](references/model-selection.md) and
+[fallback-policy.md](references/fallback-policy.md). Use the chosen model through
+available tools. Compile prompts using the H3 base or reference guide as appropriate.
+Remove storyboard-only labels, arrows, panel borders, and sketch styling from
+rendering inputs. Tie each clip to the current scene and character references.
+
+Read [qc-checklist.md](references/qc-checklist.md) for assembly and final review.
+Keep the planned shot order and audio continuity. Use a continuous score when
+requested, duck it under dialogue/SFX, and do not add subtitles unasked. Deliver
+the actual film or the requested planning package, with reproducible prompts and
+any unresolved identity, scene, or timing defects.

.agents/skills/3d-animation-short-generator/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "3D Animation Shorts"
+  short_description: "Plan stories, character anchors, shots, and edits"
+  default_prompt: "Use $3d-animation-short-generator to prepare the requested H3 prompts or video with the available tools."

.agents/skills/3d-animation-short-generator/references/fallback-policy.md

@@ -0,0 +1,21 @@
+# Targeted generation recovery
+
+Compare the failed result against the shot's actual identity/spatial anchors.
+Classify the failure: wrong input or API mapping, reference drift, overloaded
+action, timing, audio, or a render/export failure. Fix transport/configuration
+errors before spending another generation attempt.
+
+For drift, quote the relevant anchors and remove unrelated reference roles. For
+overloaded action, simplify or split the shot while preserving the requested total
+duration and handoffs. For a scene change, use the new scene reference and avoid
+conditioning on a previous scene when that would contradict the new setting.
+
+Use the retry/resource limit from the shared portable workflow. A persistent
+failure should lead to a concrete choice (simpler action, a supplied reference,
+an authorized alternate model, or an explicitly incomplete shot), not automatic
+provider switching or an endless loop. Never present a placeholder as a finished
+clip or hide an incorrect scene in final assembly.
+
+For optional pencil previews, reduce annotation or panel count. Use the text
+storyboard if visualization is not essential; do not rerender correct video
+solely because a human-review sketch changed.

.agents/skills/3d-animation-short-generator/references/model-selection.md

@@ -0,0 +1,21 @@
+# Model and format selection
+
+Honor an existing model choice. Otherwise prefer MiniMax H3 when available, and
+inspect the running backend for partitions, reference modes, audio, resolutions,
+duration limits, and long-video options. Do not assume hosted and local services
+have identical limits, quality, prices, or parameter names.
+
+If another model is requested, use its real documented input contract. A provider
+switch needs the user's authorization; never silently use a different service
+because a preset names it as a fallback. For an authorized mixed-model workflow,
+record model and settings per shot and check cross-shot style differences.
+
+Keep delivery resolution separate from native generation resolution. Record actual
+width/height, fps, frame count, and any crop, retiming, or upscaling. Do not claim
+native 2K when the video was generated at 768p and restored later.
+
+Compile shot action, camera, identity, scene, and audio into the chosen model's
+prompt format. For H3 use the bundled writing guides. Strip production labels and
+preview grids. Prompt for designed silhouettes, elastic action, material and
+lighting details when these belong to the user's 3D style; do not add text/UI just
+because a model can generate them.

.agents/skills/3d-animation-short-generator/references/qc-checklist.md

@@ -0,0 +1,19 @@
+# Assembly and review
+
+Use current clip revisions in shot-table order. Preserve useful dialogue and SFX;
+mix a continuous requested score underneath them. Inspect edits at sound bridges
+and dialogue boundaries, avoiding cut vowels and accidental audio gaps. Only add
+subtitles, music, or retiming consistent with the brief.
+
+Review identity, landmarks, screen direction, light/time of day, prop state,
+emotional payoff, intelligibility, SFX timing, and music balance. Look for leaked
+storyboard panels, sketch lines, arrows, labels, timecodes, or watermarks. Verify
+the complete file's duration, dimensions, fps, frame count, audio, and decode.
+
+Keep planning documents, references, raw clips, and final edits in distinguishable
+folders. Record the exact assets used in the final edit. After a character, scene,
+prompt, clip, or score is revised, update its downstream dependents and archive the
+previous result; do not silently combine stale and current versions.
+
+Report observed defects and what was repaired, including external editing or
+upscaling. A successful API response proves completion, not visual correctness.

.agents/skills/3d-animation-short-generator/references/shot-table-spec.md

@@ -0,0 +1,30 @@
+# Six-column shot table
+
+| Shot ID & Duration | Continuity Handoff | Reference Anchors (Spatial + Identity) | Hook Type | Shot Description (Per-Second Directives) | Audio & Dialogue Track |
+| --- | --- | --- | --- | --- | --- |
+
+Use stable IDs such as S01. Record incoming and outgoing image state, prop
+position, eyeline, motion, and emotional state in Continuity Handoff.
+
+Reference Anchors contains named landmarks and screen-relative positions; each
+character's screen position, facing and pose; recently exited characters' locations;
+lighting baseline and modifiers; and exact character/scene asset paths or IDs.
+An explicit scene/time change can replace these anchors; an accidental flip cannot.
+
+Hook Type names the shot's purpose: setup, reveal, reversal, joke, suspense,
+tenderness, chase, or callback. Do not force a joke or cut into a quiet scene.
+
+For each second or meaningful sub-second beat, cover action/pose/expression,
+camera, spatial/prop state, sound, and the handoff to the next interval. Use compact
+phrases rather than inventing movement in intentionally still moments. Include
+anticipation, squash/stretch, overshoot, and follow-through where appropriate.
+
+Keep the complete dialogue/narration script in Audio & Dialogue Track, with
+speaker, timing, delivery, and SFX. Distinguish off-screen narration from speaking
+characters; a visible listener should not lip-sync the narrator's words.
+
+Before rendering, check full time coverage, supported generation durations, a
+readable number of active characters, matching scene/identity references, and
+coherent row-to-row handoffs. Mark intentional cuts/time jumps. Fix contradictory
+states rather than using a generic continuity assurance. A visual shot may span
+multiple generation windows when the verified backend supports continuation.

.agents/skills/3d-animation-short-generator/references/storyboard-guidelines.md

@@ -0,0 +1,29 @@
+# Text storyboard and optional visualization
+
+Maintain one `storyboard.md` with a section per shot. Each section contains:
+
+1. Shot ID, title, duration, and narrative purpose.
+2. Exact scene and character reference IDs.
+3. Spatial anchors: landmarks, character positions/facing, off-screen characters,
+   key/fill/rim lighting, and important props.
+4. Incoming and outgoing continuity state.
+5. Time-ordered beats: pose/expression, camera, audio, and anchor changes.
+
+Use per-second or finer intervals only where meaningful; cover the complete shot.
+For narrated moments, specify whether a visible character speaks or only reacts.
+Optional ASCII staging diagrams can clarify screen direction but are not images
+to supply as rendering references.
+
+For heavy revision of a shot, extract its section to a named file and link it from
+the main document. Choose one authoritative version; reintegrate or update the
+link after revision so later generation cannot read stale content.
+
+Only generate pencil storyboards when requested or already authorized as previews.
+Use equal-sized time-ordered panels, with timecode, pose/expression, camera arrow,
+and audio/anchor notes. Suggested layouts: three panels in a strip, four in a 2×2
+grid, six in 2×3. More complex scenes may need fewer representative key poses.
+Keep shot/character/scene labels readable for human review, but never feed these
+annotated grids as clean final frames. The text timeline remains authoritative.
+
+If preview panels fail, simplify annotation or rely on text within the agreed
+scope. An optional visualization failure need not block a usable text storyboard.

.agents/skills/H3-SOURCES.md

@@ -0,0 +1,53 @@
+# MiniMax H3 skill provenance
+
+Upstream: https://github.com/MiniMax-AI/MiniMax-H3
+
+Source snapshot: `d21241f0a4b3acbb34c97dae47fa417b7065e438`, retrieved 2026-09-19.
+
+Source directory:
+https://github.com/MiniMax-AI/MiniMax-H3/tree/d21241f0a4b3acbb34c97dae47fa417b7065e438/skills
+
+The nine upstream skill names are retained. Each local skill corresponds to the
+same-named upstream `skills/<name>/` directory. Credit for the original creative
+workflows and templates belongs to MiniMax-AI and the upstream contributors.
+
+## Files and adaptation
+
+- `h3-prompt-writing/references/base-format.md` and `ref-format.md` are new,
+  condensed descriptions of the public prompt interface. They preserve necessary
+  field names and notation but do not copy the upstream guides or their examples.
+  The entrypoint adds a portable execution reference and clarifies clip/window
+  duration for verified long-video backends.
+- The eight style-specific `SKILL.md` files are rewritten portable adaptations of
+  the upstream workflows. Their reference documents retain the relevant creative
+  methods while removing host-specific execution assumptions.
+- The two `co-op-game-intro-generator/references/` templates retain the upstream
+  creative structure with an added portable contract and explicit user overrides.
+  Their instructions are normalized to English, and the video template expresses
+  visible UI copy as localized variables with English defaults. It also removes
+  an unrequested gender restriction and a fixed yellow/black environment
+  transformation.
+- The five 3D animation references are condensed adaptations of their upstream
+  counterparts: shot table, storyboards, model selection, fallback, and review.
+- Product, papercraft, collage, and music references extract reusable direction
+  from their corresponding upstream skill. The shared portable workflow, catalog,
+  and UI metadata are additions for this repository.
+
+No upstream `LICENSE`, `COPYING`, or `NOTICE` file was present in the Git tree at
+the pinned snapshot. Its README links to the MiniMax H3 Community License on the
+model release. Because that license has redistribution conditions, this package
+does not include the upstream prompt guides verbatim. This provenance record
+preserves attribution and does not assert that this repository's code license
+relicenses upstream material.
+
+## Runtime changes
+
+Canvas documents become local files and asset manifests; choice-card UI becomes
+the host's normal clarification mechanism when a decision is actually missing.
+Available generation/editing tools replace proprietary dispatch calls. Previous
+user choices and authorization persist across steps. Providers, prices, resolution
+support, and long-video features are checked rather than hardcoded as guarantees.
+
+User intent overrides style presets, including instrumental/no-text work, selected
+language, final morph shape, no-cut sequences, and custom player designs. Native
+generation and external processing remain separately identified in delivery.

.agents/skills/README.md

@@ -0,0 +1,96 @@
+# Portable MiniMax H3 Skills
+
+This directory contains one general prompt-writing skill and eight creative
+skills adapted from
+[MiniMax-AI/MiniMax-H3](https://github.com/MiniMax-AI/MiniMax-H3/tree/main/skills).
+See [H3-SOURCES.md](H3-SOURCES.md) for the source revision, attribution, and
+adaptation scope.
+
+## Skill catalog
+
+| Skill | Intended use |
+| --- | --- |
+| [h3-prompt-writing](h3-prompt-writing/SKILL.md) | Structured prompts for T2VA, I2VA, FL2VA, L2VA, and Ref2VA |
+| [minimalist-product-ad-generator](minimalist-product-ad-generator/SKILL.md) | Minimal product ads, isolated product references, copy, and pacing |
+| [3d-animation-short-generator](3d-animation-short-generator/SKILL.md) | 3D narrative shorts, character and environment design, shot tables, storyboards, and editing |
+| [papercraft-stop-motion-explainer](papercraft-stop-motion-explainer/SKILL.md) | Layered paper art, puppets, mechanisms, and miniature educational scenes |
+| [brand-promo-video-generator](brand-promo-video-generator/SKILL.md) | Brand promos, verified assets and claims, product interaction, and calls to action |
+| [music-video-subtitle-generator](music-video-subtitle-generator/SKILL.md) | Music videos, instrumental visual narratives, beat planning, and optional lyric typography |
+| [co-op-game-intro-generator](co-op-game-intro-generator/SKILL.md) | Two-player game menus, confirmation images, and animated world entry |
+| [paper-collage-explainer-generator](paper-collage-explainer-generator/SKILL.md) | Halftone photo collage, incremental paper assembly, and visual metaphors |
+| [handdrawn-live-video-generator](handdrawn-live-video-generator/SKILL.md) | Live action with rough luminous drawings, physical contact, continuous morphing, and delayed pursuit |
+
+## Usage
+
+Codex discovers repository skills from `.agents/skills/`. Clients that scan
+skills only when a session starts may need a new session. Claude Code reads the
+same files through the same-named relative symlinks in `.claude/skills/`. Other
+agents that support Markdown skills can load the relevant `SKILL.md` directly
+and follow its links to supporting references.
+
+Examples:
+
+```text
+Use $music-video-subtitle-generator to design a 30-second two-dimensional
+fluid-motion graphic for this instrumental track. Do not add lyrics or
+subtitles. Preserve leftward motion and initially return only a storyboard and
+three H3 prompts.
+```
+
+```text
+Use $co-op-game-intro-generator to create a two-player game intro. The players
+are Lin and Mei, the style is layered paper sculpture, and the title is
+Together. Create the confirmation image first and wait for my approval before
+generating video.
+```
+
+## MV smoke test
+
+Test `music-video-subtitle-generator` together with `h3-prompt-writing` using a
+user-owned song with timestamped lyrics and an original adult-character
+reference image:
+
+```text
+Use $music-video-subtitle-generator and $h3-prompt-writing to turn this song
+into a historical-romance music video about a woman and a man. Intercut
+narrative acting with the woman's singing performance, do not display lyric
+text, split the work according to the backend's verified clip duration, and
+preserve the complete original song in the final master.
+```
+
+Verify that the global timeline matches the source audio duration, lyrics remain
+verbatim, and each generation window records its incoming state, outgoing state,
+and local timing. Ref2VA prompts must preserve the six-section field order and
+resolve every reference label to a real input. The assembled video must decode
+completely, and its final audio must come from the supplied source rather than a
+model reconstruction.
+
+When moving the collection to another repository or a user-level skill folder,
+copy all nine skill directories and preserve their sibling layout. The eight
+creative skills share `h3-prompt-writing/references/portable-workflow.md` and the
+H3 format guides, so copying only one `SKILL.md` is insufficient. Keep each
+skill's `references/` and `agents/` directories with it. `agents/openai.yaml`
+provides optional Codex UI metadata and does not restrict the workflow to an
+OpenAI-based agent or require a dedicated plugin.
+
+## Execution boundaries
+
+Prompt writing, storyboarding, and planning require only an agent that can read
+and write files. Actual image, video, speech, music, and editing work uses tools
+or services available in the current environment. When a capability is missing,
+deliver the complete usable production package and identify the unfinished media
+step.
+
+The user's style, language, assets, audio, model selection, and existing
+authorization take priority. An MV template does not add lyrics to an
+instrumental track, and a clean motion-graphics request does not inherit film
+grain. A prompt-only request does not start generation. When the user requests
+approval of a confirmation image before rendering, preserve that checkpoint.
+
+Treat vLLM-Omni long-video extensions as capabilities of the checked-out branch
+and active backend. Do not present a development branch's continuation parameters
+as a universal H3 API. Preserve real requests and raw outputs, distinguish model
+results from post-production and upscaling, and inspect actual media properties.
+
+These skills contain no model weights and do not automatically start a service
+or invoke a paid generation endpoint.

.agents/skills/brand-promo-video-generator/SKILL.md

@@ -0,0 +1,66 @@
+---
+name: brand-promo-video-generator
+description: Plan, prompt, or produce brand, product, website, app, and service promotional shorts using traceable assets, accurate claims, timed beats, and clear calls to action. Use for launch and campaign videos, not invented product evidence or long narrative films.
+metadata:
+  compatibility: File-based workflow with sibling h3-prompt-writing and available research, media generation, inspection, and editing tools.
+---
+
+# Brand Promo Video Generator
+
+Read the [portable workflow](../h3-prompt-writing/references/portable-workflow.md).
+Use supplied brand assets, source links, and existing session decisions. No
+proprietary tool allowlist or canvas is required. A completed promo requires real
+generation/editing capability; otherwise deliver an explicitly incomplete production
+package with finished prompts and a concrete generation handoff.
+
+## Establish facts and assets
+
+Inspect product images, UI screenshots, logo exports, typography, colors, and
+campaign material. Resolve duration, ratio, audience, campaign focus, exact claims,
+CTA, copy language, narration, and required deliverable from context. Ask for
+missing identity-bearing assets when they are essential; do not fabricate logos,
+product UI, screenshots, results, or performance claims.
+
+Build a compact truth sheet from user-supplied originals and official sources.
+Record asset ID, role, local path, source URL/note, and verification status. Do not
+confuse a retrieved search thumbnail with an original brand asset. Distinguish
+verified facts from clearly labeled concept copy. Preserve the requested language;
+chat language alone does not justify translating supplied campaign text.
+
+## Design a product-specific story
+
+Choose a spine that exposes real cause and effect:
+
+- Software: user intent → actual interaction → useful output → evidence → CTA.
+- Physical product: reveal → interaction → mechanism/detail → use → benefit → CTA.
+- Service: situation → process → evidence → outcome → promise.
+- Image-led brand: authentic imagery → repeated visual motif → emotional payoff.
+
+Plan time/frame ranges, asset IDs, one main action, product evidence, exact copy,
+readable hold, color state, camera, transition, and rhythm. Use the actual output
+fps for frame timing. Five to eight major beats can fit a 15-second promo, but
+readability takes priority over a fixed beat count.
+
+Let product edges, UI movement, light, cursor paths, or matched geometry motivate
+transitions. Use a few energy peaks and deliberate settling moments. Preserve logo
+clear space and readable screenshots. Avoid invented HUDs or metrics that look like
+product proof. Summarize the truth sheet and plan before rendering; continue under
+existing authorization unless the user requested approval at that stage.
+
+## Produce and inspect
+
+Use authentic still/UI plates for exact content. Generated material can supply
+conceptual scenes, atmosphere, and transition geometry with clear provenance.
+Compile actual H3 references and shots using the appropriate bundled guide.
+Check backend resolution, duration, and audio support before calling it.
+
+Use native instrumental/SFX direction when appropriate; do not generate a second
+score automatically. Separate narration or music is useful when requested for
+control or remixing. Preserve original SFX during assembly when they matter, and
+duck music under speech. Do not add subtitles unless requested.
+
+Check claims and exact product names against sources, inspect logo geometry and
+copy, verify real UI states, and review transitions, aspect ratio, duration, audio,
+and decoding. For wrong identity-bearing content, use the correct supplied/source
+asset rather than polishing an imitation. Deliver the video, prompts/timeline,
+asset manifest, and any remaining factual or visual limitations.

.agents/skills/brand-promo-video-generator/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "Brand Promo Videos"
+  short_description: "Build brand promos from verified assets and claims"
+  default_prompt: "Use $brand-promo-video-generator to prepare the requested H3 prompts or video with the available tools."

.agents/skills/co-op-game-intro-generator/SKILL.md

@@ -0,0 +1,73 @@
+---
+name: co-op-game-intro-generator
+description: Design or generate a two-player co-op game menu and opening animation from player names, title, style, and optional identity references. Use for character-led game intros, not playable game implementation or complex multi-page interfaces.
+metadata:
+  compatibility: File-based workflow with sibling h3-prompt-writing and bundled image/video templates. Rendering uses available generation tools.
+---
+
+# Co-op Game Intro Generator
+
+Read the [portable workflow](../h3-prompt-writing/references/portable-workflow.md).
+Before writing a confirmation-image prompt, read
+[h3-confirmation-image-template.md](references/h3-confirmation-image-template.md).
+Before writing the video prompt, read
+[h3-video-prompt-template.md](references/h3-video-prompt-template.md).
+Both templates are required package resources, not remote tools.
+
+## Resolve player and style bindings
+
+Reuse or collect the two player names, game title, visual style, desired duration,
+ratio, and optional identity references. The template defaults to a 15-second,
+16:9 menu-to-world sequence; adapt it to explicit user requirements. Do not assume
+the players' gender, body shape, outfits, or equipment from a generic example.
+
+Use each portrait only for recognizable identity anchors: face silhouette, hair,
+glasses, proportions, and distinguishing traits. Reinterpret rendering, skin
+texture, lighting, and costume in the selected visual style. If a portrait or body
+comparison reference is absent, write a text design rather than inventing an asset
+label. Maintain player-to-name mapping across all references and frames.
+
+## Confirmation image
+
+Fill the template's style, palette, characters, title, and name variables. Keep
+the two characters central, player cards toward the upper left, menu on the right,
+and the continue action as the main interaction focus. The title establishes
+identity without competing with that focus. Buttons, icons, typography, and all
+player/equipment panels share one palette and style. Keep menu labels on one line.
+
+The style changes material, shapes, fonts, and decoration; it should not randomly
+rearrange the menu hierarchy. User-requested layout or ratio changes take priority
+over template defaults. Preserve the exact supplied title/names and requested copy.
+
+For image-only/prompt-only work, stop at that deliverable. For video production,
+generate and inspect the confirmation image if within scope. Honor a request to
+approve the image first; otherwise use the established end-to-end authorization
+and continue after reporting the selected image and fixing material defects.
+
+## Video event plan
+
+Fill the video template from the current image and player bindings. Its baseline
+sequence is menu → first player's light equipment → second player's heavy equipment
+→ shared confirmation → world loading → both enter the world. Equipment, world,
+body contrast, and timing are editable design choices; preserve requested identity
+and style instead of forcing a cyberpunk example onto another setting.
+
+Bind UI reference, each actual portrait, and any real body-comparison image to
+distinct roles. Compile this creative template into the H3 mode guide's fields;
+the mixed-language template is not itself an API request. Preserve literal visible
+UI text in its required language, and remove all unfilled variables before dispatch.
+Do not treat a composition reference as an exact first-frame anchor unless the
+selected backend actually supports that role.
+
+## Review and repair
+
+Check both names, exact menu strings, layout, player identity, side assignment,
+equipment distinction, and the transition into the requested world. Match UI
+hover/click, mechanical locks, loading, and footsteps to the visible events.
+For identity swaps, strengthen name/position/reference bindings; for weak style,
+revise palette, material, character, and typography together. Reduce excessive
+UI text if necessary without silently changing required player names.
+
+Deliver the requested prompts/assets or the inspected intro video, with reference
+roles and any text/timing limitations. Missing templates block template execution;
+missing generation capability still permits a complete prompt handoff.

.agents/skills/co-op-game-intro-generator/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "Co-op Game Intros"
+  short_description: "Design two-player menus and world-entry animation"
+  default_prompt: "Use $co-op-game-intro-generator to prepare the requested H3 prompts or video with the available tools."

.agents/skills/co-op-game-intro-generator/references/h3-confirmation-image-template.md

@@ -0,0 +1,397 @@
+# H3 Confirmation Image Template
+
+## Portable adaptation
+
+This is a creative template, not a tool call. User requirements override its
+default ratio, layout, poses, and wording. Fill every relevant variable from the
+brief; set `{character_count}` to two for this skill. When portraits are absent,
+replace portrait references with explicit character designs. Do not claim to
+reference an image that was not supplied. Use the image-generation capability
+available in the host environment; no specific image provider is required.
+
+## Core Rule
+Generate a polished 16:9 landscape confirmation image for a cooperative game main menu.
+The image consists of two independent systems:
+• UI Framework (fixed)
+• Visual Style (dynamic)
+The UI Framework defines the layout, composition, information hierarchy, interaction logic, reading flow, menu structure, and readability. These elements must remain unchanged.
+{visual_style} only controls the artistic appearance, including rendering medium, materials, color language, lighting mood, environment style, character design, costume language, UI materials, icon language, typography appearance, decorative motifs, and overall atmosphere.
+Changing {visual_style} should feel like applying a different art direction to the same professionally designed game menu, never redesigning the interface itself.
+If any default description conflicts with {visual_style}, prioritize {visual_style} while preserving the UI framework.
+
+---
+
+## Design Goal
+Create a premium commercial console game main menu mockup in 16:9 landscape format. The image should combine the quality of a polished game promotional artwork with the clarity of a professional game UI. Characters and interface must feel naturally integrated, with strong visual hierarchy, excellent readability, clean composition, and a clear interaction focus.
+
+---
+
+## Overall Style
+Follow the user-selected visual style: {visual_style}.
+The selected style controls the rendering medium, illustration style, materials, palette language, environment, character appearance, outfit design, UI appearance, icon language, typography treatment, decorative elements, and atmosphere.
+All visual elements must belong to one unified artistic language.
+Do not mix incompatible visual styles within the same image.
+The UI framework, layout, hierarchy, composition, and interaction logic must remain consistent regardless of style.
+
+---
+
+## Color Palette
+Derive the entire color system from {visual_style}.
+Use:
+• {main_background_color} as the primary background color.
+• {ui_body_color} as the main UI color.
+• {text_color} as the primary text color.
+• {functional_accent_color} as the interaction highlight color.
+• Red only for warning, danger, or exit actions.
+Limit the overall palette to five primary colors.
+All buttons, player cards, icons, borders, highlights, decorative elements, typography, and visual effects must follow the same color system.
+Avoid introducing unrelated colors outside the selected palette.
+
+---
+
+## Composition
+Use a fixed 16:9 landscape layout.
+Place {character_count} playable characters in the center as the visual subject.
+The UI surrounds the characters without blocking them.
+Place the player information cards in the upper-left corner.
+Place the main menu vertically on the right.
+Place a decorative horizontal strip along the bottom, adapting naturally to {visual_style} (such as ribbon, warning strip, wooden plank, stone slab, energy bar, scroll, etc.).
+Keep a clear Z-shaped reading flow:
+Player Cards → Characters → Menu → Continue Button.
+The Continue button must always remain the primary visual focus.
+Maintain generous negative space and excellent readability.
+
+---
+
+## Background
+Generate a background naturally derived from {visual_style}.
+The background supports the interface rather than competing with it.
+Keep enough clean negative space behind all important UI elements.
+Use subtle textures, environmental details, and decorative motifs only where appropriate for the selected style.
+Avoid excessive visual noise, dense decoration, heavy dirt, scratches, paint splatter, or distracting textures.
+The overall background should feel polished, premium, modern, and suitable for a commercial game menu.
+
+---
+
+## Character Style
+All characters strictly follow {visual_style}.
+The selected style determines:
+• rendering medium
+• body proportion
+• facial stylization
+• costume language
+• material treatment
+• lighting response
+• animation style
+• overall character quality
+Only the character identity anchors and pose remain fixed. Facial rendering, eye design, nose simplification, mouth shape, skin treatment, expression style, and head proportion must be fully reinterpreted in {visual_style}.
+No default character style may override {visual_style}.
+
+---
+
+## Character A / PLAYER 1
+### Identity
+Use character reference image 1 only for identity mapping.
+The reference image donates identity anchors only: recognizable face silhouette, hairstyle, glasses if present, relative facial proportions, and distinctive personal traits.
+It does not donate photographic realism, real skin texture, real-world lighting, camera quality, or the original image style.
+Do not lose the character's identity anchors, but the face must be redesigned into {visual_style}.
+All facial rendering, eye design, nose simplification, mouth shape, skin treatment, expression style, and head proportion must follow {visual_style}.
+Do not swap identity with other characters.
+
+### Appearance
+Character rendering, expression, outfit, materials, and accessories are fully derived from {visual_style} while maintaining a recognizable playable-game-character silhouette.
+
+### Action
+Character A sits cross-legged on the ground.
+Both hands support the body on the floor.
+The upper body leans slightly backward.
+Character A looks upward toward the camera.
+
+---
+
+## Character B / PLAYER 2
+### Identity
+Use character reference image 2 only for identity mapping.
+The reference image donates identity anchors only: recognizable face silhouette, hairstyle, glasses if present, relative facial proportions, and distinctive personal traits.
+It does not donate photographic realism, real skin texture, real-world lighting, camera quality, or the original image style.
+Do not lose the character's identity anchors, but the face must be redesigned into {visual_style}.
+All facial rendering, eye design, nose simplification, mouth shape, skin treatment, expression style, and head proportion must follow {visual_style}.
+Do not swap identity with other characters.
+
+### Appearance
+Character rendering, expression, outfit, materials, and accessories are fully derived from {visual_style} while maintaining a recognizable playable-game-character silhouette.
+
+### Action
+Character B sits cross-legged on the ground.
+Both hands rest naturally in front of the legs.
+The upper body leans slightly forward.
+Character B looks directly toward the camera.
+
+---
+
+## Lighting
+Lighting follows {visual_style} while preserving clear readability.
+Use a primary top light, secondary fill light, soft ambient bounce, contact shadows, gentle rim lighting, and high-quality global illumination where appropriate.
+Faces should remain clearly visible.
+Characters must integrate naturally into the environment while maintaining strong separation from the background.
+
+---
+
+## Game UI
+Create a professional commercial console game main menu.
+The menu structure, interaction hierarchy, spacing, composition, reading flow, and information architecture remain fixed.
+{visual_style} only changes the artistic appearance of the interface.
+The selected style may affect:
+• UI materials
+• surface textures
+• border treatment
+• corner shape
+• decorative motifs
+• color rendering
+• lighting treatment
+• icon language
+• typography appearance
+• interaction effects
+It must NOT change:
+• menu hierarchy
+• button order
+• button placement
+• player-card placement
+• composition
+• interaction logic
+• information architecture
+• readability
+The entire interface should feel like one professionally designed game UI system rather than separate elements assembled together.
+
+---
+
+## UI Visual Language
+Derive the complete UI appearance from {visual_style}.
+The selected style determines:
+• button materials
+• border language
+• panel materials
+• decorative details
+• edge treatment
+• visual weight
+• highlight style
+• interaction feedback
+• shadow treatment
+• texture language
+Possible materials include wood, stone, paper, leather, cloth, glass, crystal, hologram, neon, metal, ceramic, carved surfaces, painted surfaces, magical energy, pixel blocks, watercolor paper, ink brush, comic graphics or any other material naturally belonging to {visual_style}.
+Every UI element must share the same material language and artistic style.
+
+---
+
+## Button Layout Rules
+Place all menu buttons vertically on the right side.
+Maintain a consistent vertical rhythm.
+Every button uses the same design family.
+Button width automatically adapts to text length.
+Keep all menu titles on a single line.
+No automatic wrapping.
+No stacked text.
+Text remains horizontally centered.
+Keep consistent padding.
+Maintain equal spacing between buttons.
+Avoid visual clutter.
+
+---
+
+## Continue Button
+The Continue button is always the primary visual focus.
+Make it visually dominant using the emphasis method naturally belonging to {visual_style}.
+Possible emphasis methods include:
+• larger size
+• stronger contrast
+• brighter color
+• unique material
+• distinctive border
+• lighting
+• glow
+• engraving
+• magical energy
+• holographic effect
+• embossed frame
+• animated visual cue
+Do not force glow or neon if they conflict with {visual_style}.
+The Continue button should immediately attract the player's attention.
+
+---
+
+## Start New Game Button
+Place above Continue.
+Use the standard secondary button appearance.
+Keep the visual language consistent with the selected style.
+Add one small style-appropriate icon if suitable.
+Maintain single-line text.
+
+---
+
+## Settings Button
+Use the standard secondary button appearance.
+Use one settings icon naturally derived from {visual_style}.
+Maintain consistent proportions, spacing, and hierarchy.
+
+---
+
+## Exit Game Button
+Use the standard secondary button appearance.
+Use red only as the danger indicator.
+Red should remain an accent rather than the dominant color.
+Add one style-appropriate exit icon if suitable.
+
+---
+
+## Player Cards
+Create an upper-left player information system for {character_count} players.
+The appearance follows {visual_style}.
+Card materials, borders, corners, textures, shadows, decorations, and icons are automatically derived from the selected style.
+Each player card contains:
+• Player label
+• Player nickname
+• READY status
+Player names:
+PLAYER 1
+{player1_name}
+PLAYER 2
+{player2_name}
+READY always uses the functional accent color.
+The cards should remain highly readable and visually connected to the rest of the interface.
+
+---
+
+## Icon System
+Derive all icons from {visual_style}.
+Use one unified icon family throughout the entire interface.
+Possible icon languages include:
+• minimal outline
+• solid filled
+• geometric
+• comic
+• pixel
+• engraved
+• carved
+• painted
+• brush
+• rune
+• glyph
+• holographic
+• crystal
+• paper-cut
+• origami
+• embossed
+• low-poly
+• flat design
+• stylized 3D
+All icons must share:
+• the same artistic style
+• the same material language
+• the same perspective
+• the same lighting treatment
+• the same stroke language
+• the same color hierarchy
+Icons should remain simple, recognizable, and functional.
+Each interface block may contain at most one horizontal row of icons.
+No wrapping.
+No vertical stacking.
+Icons should support the interface rather than dominate it.
+
+---
+
+## Typography
+Typography hierarchy remains fixed.
+Typography appearance follows {visual_style}.
+The selected style controls:
+• font family
+• decorative treatment
+• texture
+• material
+• outline
+• lighting
+• shadow
+• wear pattern
+Maintain:
+• excellent readability
+• strong visual hierarchy
+• single-line menu buttons
+• consistent spacing
+• high contrast
+Button width should always adapt to text length.
+Never allow menu titles to wrap onto multiple lines.
+
+---
+
+## Game Title
+Display the game title:
+{game_title}
+The title should become the primary branding element of the menu.
+Its material, decorative style, typography, and rendering all follow {visual_style}.
+Avoid imitating existing commercial game logos.
+Create an original title treatment inspired by the selected style.
+Keep the title highly readable.
+Avoid random symbols or meaningless letters.
+
+---
+
+## Interaction States
+Represent the interface as an active game menu.
+Buttons may display appropriate interaction feedback naturally belonging to {visual_style}.
+Possible interaction feedback includes:
+• hover
+• pressed state
+• magical glow
+• energy pulse
+• holographic highlight
+• carved illumination
+• ink spread
+• brush bloom
+• metallic reflection
+• paper emboss
+• animated border
+• pixel flashing
+Only use interaction effects that naturally belong to the selected style.
+Avoid forcing modern UI effects into incompatible artistic styles.
+
+---
+
+## Rendering
+Generate a polished AAA-quality commercial game menu artwork.
+Characters, environment, UI, typography, icons, and decorative elements should feel like one cohesive production.
+Maintain premium rendering quality.
+Strong visual hierarchy.
+Professional commercial presentation.
+High readability.
+Balanced composition.
+Clean finish.
+
+---
+
+## Negative Constraints
+No split screen.
+No duplicated characters.
+No identity swap.
+No loss of identity anchors. No photorealistic face unless {visual_style} explicitly requires photorealism. No realistic photo face pasted onto a stylized body. No mismatched face style.
+No hairstyle changes.
+No unreadable UI.
+No distorted buttons.
+No overlapping interface.
+No random text.
+No random symbols.
+No watermark.
+No copied game logo.
+No branded interface.
+No inconsistent icon styles.
+No mixed UI materials.
+No excessive decoration.
+No overwhelming background.
+No visual clutter.
+No broken layout.
+No low readability.
+
+---
+
+## Medium Lock
+Render as a polished commercial game key art combined with a professional console game main menu mockup.
+The interface remains fixed.
+Only the artistic appearance changes according to {visual_style}.

.agents/skills/co-op-game-intro-generator/references/h3-video-prompt-template.md

@@ -0,0 +1,249 @@
+# H3 Video Prompt Template
+
+## Portable adaptation
+
+This is a creative planning template. User requirements override its default
+15-second timeline, body proportions, equipment, environment, ratio, and palette.
+Retiming must preserve readable event order. Use only supplied reference assets:
+omit `{height_ref}` or identity-image bindings when absent and replace their role
+with text design constraints. Compile the filled plan into the H3 base or Ref2VA
+format before rendering; preserve exact visible UI strings in the user's language.
+
+Use available tools and the actual backend's parameter schema. Audio directions
+describe desired content rather than asserting that a particular API supports it.
+
+## Template completion
+
+Resolve the palette variables `{background_color}`, `{ui_color}`, `{text_color}`,
+`{functional_accent_color}`, and `{power_accent_color}` from the selected style.
+Keep the five values distinct enough for readable UI contrast. Use the power
+accent for danger, exit, and warning states unless the user supplies a separate
+semantic-color requirement.
+
+Fill visible copy in the user's requested language. The following English values
+are defaults only, not mandatory output:
+
+- `{player1_label}` / `{player2_label}`: `PLAYER 1` / `PLAYER 2`
+- `{ready_label}`: `READY`
+- `{start_label}`, `{continue_label}`, `{settings_label}`, `{exit_label}`:
+  `START NEW GAME`, `CONTINUE`, `SETTINGS`, `EXIT GAME`
+- `{player1_equipment_label}`: `RIGHT ARM EQUIPMENT`
+- `{player1_option_a}` / `{player1_option_b}`: `PHANTOM GRIP` / `CHRONOS CLAW`
+- `{player2_equipment_label}`: `ARMAMENT CUSTOMIZATION`
+- `{hand_label}`, `{forearm_label}`, `{elbow_label}`, `{upper_arm_label}`:
+  `HAND`, `FOREARM`, `ELBOW`, `UPPER ARM`
+- `{confirm_label}` / `{loading_label}`: `CONFIRM CONFIG` / `LOADING`
+
+Before compiling the rendering prompt, search the filled template for every
+remaining token enclosed in braces. Replace required values. For an absent optional
+reference, remove its binding and rewrite every dependent clause as a text-only
+constraint. Do not dispatch a prompt while any template token remains.
+
+## Prompt principle
+Use the same method as the GPT confirmation-image prompt:
+
+**Fixed video event framework + dynamic user-style fill + locked character identity + palette-linked UI system.**
+
+The video prompt must not blindly preserve the source prompt's default style. It must preserve the timeline, UI events, player positions, equipment logic, and text structure, while dynamically rewriting visual treatment, palette, character rendering, lighting, UI surface, icons, and city-world style according to the user's selected style.
+
+## Priority order
+1. User-selected style: {visual_style}
+2. Confirmed image / UI reference: {ui_ref}
+3. PLAYER 1 identity reference: {player1_ref}
+4. PLAYER 2 identity reference: {player2_ref}
+5. Height/body comparison reference: {height_ref}
+6. Fixed source video event framework
+
+## Reference roles
+- {ui_ref}: confirmed first image. Use it to lock UI layout, menu hierarchy, color system, typography scale, button structure, character-game integration, and overall composition logic.
+- {player1_ref}: PLAYER 1 identity anchor. Lock exact face, hairstyle, glasses if present, facial proportions, body identity, and nickname mapping to {player1_name}.
+- {player2_ref}: PLAYER 2 identity anchor. Lock exact face, hairstyle, facial proportions, body identity, and nickname mapping to {player2_name}.
+- {height_ref}: body comparison anchor. Lock the visible contrast between the two players and prevent identical body proportions.
+
+## Global style baseline
+The overall art direction must prioritize the user's selected {visual_style}.
+Preserve these fixed qualities: a game main-menu interface, premium game-trailer
+finish, deep integration between the UI and characters, modern commercial game
+UI design, strong visual impact, clean composition, and restrained decoration.
+Derive the remaining visual treatment from {visual_style}: character rendering,
+expressions, costume language, color system, lighting temperature, UI materials,
+button icons, typography texture, and the world revealed after loading.
+
+## Palette system
+Derive the complete video palette from {visual_style}, with linked UI colors:
+
+- Use {background_color} as the primary background and world color.
+- Use {ui_color} as the main UI color.
+- Use {text_color} as the text color.
+- Use {functional_accent_color} as PLAYER 1's functional accent color.
+- Use {power_accent_color} as PLAYER 2's power accent and for danger, exit,
+  and warning states.
+- Limit the complete palette to five colors.
+- Use a vivid, modern, high-contrast color language consistent with
+  {visual_style}.
+
+Menus, equipment panels, player cards, buttons, the HUD, loading bars, icons,
+and text must share this color system. Do not introduce unrelated colors.
+
+## Character identity and style lock
+PLAYER 1 follows the facial identity in {player1_ref}. Preserve the face,
+facial proportions, hairstyle, glasses when present, personal identity, and the
+mapping to the nickname {player1_name}. Adapt expression, rendering, clothing,
+and mechanical equipment to {visual_style}. PLAYER 1 always stays on the left,
+with a taller, slender, agile build. Equipment uses the functional accent color,
+and the mechanical claw is light, narrow, and flexible.
+
+PLAYER 2 follows the facial identity in {player2_ref}. Preserve the face,
+facial proportions, hairstyle, personal identity, and the mapping to the
+nickname {player2_name}. Adapt expression, rendering, clothing, and mechanical
+equipment to {visual_style}. PLAYER 2 always stays on the right, with a shorter,
+broader, power-oriented build. Equipment uses {power_accent_color}, and the
+mechanical fist is broad, heavy, and visibly weighty.
+
+Do not swap character identities, merge faces, swap nicknames, or converge the
+two body types.
+
+## Fixed timeline framework
+
+### [0-2 seconds] - Two-player main menu
+
+Composition and camera: A high-angle extreme wide shot follows the composition
+logic of {ui_ref}. The camera tilts slightly downward and slowly pushes in.
+
+Visuals: PLAYER 1 ({player1_name}) and PLAYER 2 ({player2_name}) sit side by
+side at the center. PLAYER 1 remains on the left and PLAYER 2 on the right. They
+look up toward the camera with only natural breathing, blinking, and subtle body
+movement.
+
+UI: The upper-left player card displays `{player1_label}`, `{player1_name}`, and
+`{ready_label}` exactly as filled. The second card in the upper-center or
+upper-left two-card system displays `{player2_label}`, `{player2_name}`, and
+`{ready_label}` exactly as filled. The vertical menu on the right displays
+`{start_label}`, `{continue_label}`, `{settings_label}`, and `{exit_label}`
+exactly as filled. `{continue_label}` is the visual center and primary
+highlighted button.
+
+Dynamic style fill: Adapt the menu background, ground texture, button shapes,
+icons, typography, borders, glow, and sticker treatment to {visual_style}, while
+preserving the layout and hierarchy in {ui_ref}.
+
+Sound: Menu ambience, subtle UI-hover sounds, and a low electronic atmosphere
+before the click. Ignore this direction when generating silent video.
+
+### [2-4 seconds] - PLAYER 1 right-arm configuration
+
+Composition and camera: In a medium shot, the camera pushes smoothly from the
+main menu toward PLAYER 1's right arm. PLAYER 2 remains visible and stable in the
+background.
+
+UI: The right-side menu contracts and slides away. A panel with functional-accent
+guide lines slides in from the left and displays `{player1_label}` and
+`{player1_equipment_label}` exactly as filled. The equipment list first
+highlights `{player1_option_a}`, then moves the selection to
+`{player1_option_b}`.
+
+Action: PLAYER 1's right cuff opens automatically and a lightweight mechanism
+unfolds beneath the forearm. The fingers spread as long claw-like joints slide
+into position and lock one by one, briefly revealing fine wiring, miniature
+pistons, and metal connectors. Functional-accent LEDs illuminate sequentially
+when configuration completes.
+
+Dynamic style fill: Adapt the mechanism, panel material, icons, lines, locking
+animation, and light effects to {visual_style}. Keep the result light, precise,
+and flexible, proportioned to PLAYER 1 without changing the face, hairstyle, or
+main clothing design.
+
+Sound: Precise mechanical unfolding, light UI switching tones, and small locking
+clicks.
+
+### [4-7 seconds] - PLAYER 2 heavy-arm configuration
+
+Composition and camera: A medium shot tracks smoothly between the players and
+arcs toward PLAYER 2's left side. PLAYER 1 remains in the background, quietly
+examining the configured mechanical hand.
+
+UI: A {power_accent_color} panel slides in and displays `{player2_label}` and
+`{player2_equipment_label}` exactly as filled. Its grid contains `{hand_label}`,
+`{forearm_label}`, `{elbow_label}`, and `{upper_arm_label}`, with the selection
+moving quickly but legibly through all four.
+
+Action: PLAYER 2's left sleeve opens in sections. Heavy forearm plates spring
+outward, the previous components detach, and new armor slides into place along
+guide rails. A thick mechanical bearing replaces the elbow joint, and the broad
+mechanical hand reassembles and locks. Thick wiring, hydraulic pistons, and a
+dark metal skeleton appear briefly during the replacement. A restrained warm
+indicator illuminates as each component locks.
+
+Dynamic style fill: Adapt the heavy arm, UI panel, component icons, materials,
+and light effects to {visual_style}. Keep it broad, weighty, and powerful, in
+clear contrast with PLAYER 1's lightweight claw.
+
+Sound: Low motor movement, heavy mechanical engagement, and weighty locking
+feedback.
+
+### [7-8.5 seconds] - Shared configuration confirmation
+
+Composition and camera: Pull back smoothly to a medium two-player composition,
+with PLAYER 1 on the left and PLAYER 2 on the right.
+
+UI: The two equipment panels converge at center to form a shared button that
+displays `{confirm_label}` exactly as filled. Adapt its border, glow, icon, and
+sticker treatment to {visual_style}, while keeping the hierarchy and text legible.
+
+Action: The cursor clicks the button. A functional-accent energy pulse travels
+through PLAYER 1's mechanical claw, while a {power_accent_color} pulse travels
+through PLAYER 2's fist. All UI panels contract inward and disappear. Both
+players uncross their legs and adjust their posture: PLAYER 1 lifts one knee
+lightly and articulates the claw fingers in sequence; PLAYER 2 plants one foot
+firmly and slowly closes the heavy fist.
+
+Sound: Confirmation tone, two-color energy pulses, and UI contraction.
+
+### [8.5-10 seconds] - Shared world loading
+
+Composition and camera: A wide shot reveals a shared loading bar at the bottom.
+
+UI: The loading bar displays `{loading_label}` exactly as filled and fills
+rapidly from 0% to 100%. Its left half uses {functional_accent_color} and its
+right half uses {power_accent_color}. Adapt the HUD and loading-bar shape,
+border, texture, and typography to {visual_style}, while keeping them legible.
+
+Environment transformation: Continuously transform the confirmed
+{visual_style} menu background into a game world in the same style. Menu strips,
+color blocks, and patterns become corresponding roads, structures, and world
+elements. Preserve the confirmed palette rather than forcing a yellow-and-black
+or cyberpunk city.
+
+Key constraint: The transformation remains continuous and natural, with no hard
+cut, smoke occlusion, or identity change.
+
+Sound: Rising loading tone and an ambience transition from menu to game world.
+
+### [10-15 seconds] - Two-player world entry
+
+Composition and camera: Move from an extreme wide shot into a third-person
+tracking view. At 100% loading, both characters stand simultaneously. The camera
+descends smoothly and arcs behind them into a stable two-player cooperative view.
+
+World style: Generate the complete world from {visual_style}. Preserve the game
+opening structure: dense buildings, roads, signal lights, signs, crowds or other
+moving background elements, passing vehicles, overhead wires, industrial pipes,
+and a distant skyline. Materials, architecture, signage, lighting, and motion
+must follow {visual_style}. Do not let the source template's cyberpunk default
+override the user unless the selected style is cyberpunk.
+
+Character relationship: Clearly show the players from behind and preserve their
+body contrast. PLAYER 1 is on the left, tall and slender, with the lightweight
+claw hanging naturally. PLAYER 2 is on the right, shorter and broader, with the
+heavy fist slightly raised. PLAYER 1 steps forward first, PLAYER 2 follows, and
+they enter the street side by side.
+
+HUD: The HUD fades in. A mini-map appears in the upper right. Separate status
+bars labeled `{player1_name}` and `{player2_name}` appear in the lower left.
+The first uses {functional_accent_color}; the second uses {power_accent_color}.
+A shared objective marker appears in the street ahead.
+
+Sound: City ambience, distant vehicles, footsteps, and a HUD fade-in cue.
+
+## Negative constraints
+No extra player, duplicated character, identity or username swapping, merged bodies, unintended face/hair changes, missing player, unrequested split screen or cuts, random camera shake, floating body parts, gruesome dismemberment, unwanted weapons, style-incompatible neon, unreadable UI, random letters, misspelled usernames, extra menu options, copied game logo, or watermark. Preserve the user's chosen character designs and palette.

.agents/skills/h3-prompt-writing/SKILL.md

@@ -0,0 +1,42 @@
+---
+name: h3-prompt-writing
+description: Write MiniMax H3 video generation prompts for T2VA, I2VA, FL2VA, L2VA, and Ref2VA. Use when rewriting multimodal requests into H3 prompt structures, composing integrated_multimodal_description, overall_soundscape, and non_diegetic_music, aligning keyframes, or defining reference labels for images, videos, and audio.
+metadata:
+  compatibility: Portable to any agent that can read local files — no external API calls, MiniMax Hub tools, or proprietary runtime required. The agents/openai.yaml file only adds optional ChatGPT/Codex UI metadata; it does not restrict the skill to OpenAI agents.
+---
+
+# H3 Prompt Writing
+
+## Workflow
+
+0. Read [portable-workflow.md](references/portable-workflow.md) for scope, available tools, durable artifacts, and generation handoff. For prompt-only work, no generation tools are required.
+1. Identify the input mode: T2VA, I2VA, FL2VA, L2VA, or full-reference Ref2VA.
+2. For base text/keyframe modes, read `references/base-format.md` and follow its final prompt structure.
+3. For full-reference mode, read `references/ref-format.md` and follow its six-section rewrite format.
+4. Preserve the exact field names, section order, labels, and timing notation from the selected guide.
+
+## Base Modes
+
+- T2VA: build the full audiovisual timeline from text.
+- I2VA: start from the first frame and develop forward from it.
+- FL2VA: describe the continuous path between the first and last frames.
+- L2VA: infer a plausible opening and converge to the supplied last frame.
+
+Use `integrated_multimodal_description`, `overall_soundscape`, and `non_diegetic_music` in the order shown in `references/base-format.md`.
+
+## Full-Reference Mode
+
+Ref2VA rewrites use `subject_definitions`, `summary`, `retention_analysis`, `detailed_description`, `overall_soundscape`, and `non_diegetic_music` in that order. Reference labels stay consistent across all sections.
+
+Read `references/ref-format.md` for label rules and retention analysis.
+
+## Output Rules
+
+- Write rewrite sections in English; preserve dialogue, lyrics, and visible scene text in their original language.
+- Describe each shot by composition, subjects, environment, actions, camera, sound, and the exact point where referenced content appears.
+- Avoid plot summaries, unresolved reference labels, and timing that does not match the requested duration.
+## Tips for Better Results
+- Match the description to the effective duration of the current clip or generation window. The upstream guides describe 4–15-second clips; longer work requires a verified backend extension or an explicit assembly plan, as described in the portable workflow.
+- Keep reference labels consistent (e.g. `<Picture 1>`, `<Video 1>`, `<Audio 1>`) across every section.
+- Prefer concrete visual and audio details over abstract words like "cinematic" or "beautiful".
+- When using keyframes (I2VA / FL2VA / L2VA), clearly state how the first and/or last frame connects to the timeline.

.agents/skills/h3-prompt-writing/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "H3 Prompt Writing"
+  short_description: "Write structured H3 prompts for each input mode"
+  default_prompt: "Use $h3-prompt-writing to prepare the requested H3 prompts or video with the available tools."

.agents/skills/h3-prompt-writing/references/base-format.md

@@ -0,0 +1,86 @@
+# H3 base-mode prompt format
+
+Use this reference for T2VA, I2VA, FL2VA, and L2VA prompt writing. It is a
+compact, repository-authored description of the required output contract, not a
+copy of an upstream prompt guide.
+
+## Select the anchor mode
+
+- T2VA has no image anchor. Build the audiovisual sequence from the brief.
+- I2VA binds `<Picture 1>` to the first frame at 0.00 seconds.
+- FL2VA binds `<Picture 1>` to the first frame and `<Picture 2>` to the final
+  frame. Describe the observable path between them.
+- L2VA binds `<Picture 1>` to the final frame. Infer a compatible opening and
+  converge to the supplied image.
+
+For an image-anchored mode, put a plain-English alignment statement before the
+structured fields. State each picture label, its shot, and its exact target time.
+The ending time must match the effective clip duration, formatted to two decimal
+places. Do not invent a picture label when no image was supplied.
+
+## Emit three fields in order
+
+```text
+integrated_multimodal_description: [Shot 1] ...
+
+overall_soundscape: ...
+
+non_diegetic_music: ...
+```
+
+`integrated_multimodal_description` is the playback-ordered visual and diegetic
+audio plan. Establish the initial style, framing, subjects, environment, and
+spatial relationships, then describe actions and state changes. For keyframe
+modes, explicitly preserve or approach the relevant composition, identity,
+clothing, colors, objects, and geometry.
+
+`overall_soundscape` summarizes ambience, physical effects, and non-verbal human
+sound. Do not repeat dialogue or music already described elsewhere. Use `N/A`
+only for explicitly silent output.
+
+`non_diegetic_music` describes music heard by the audience but not the characters.
+Specify audible properties such as instrumentation, tempo, rhythm, dynamics, and
+ending behavior. Put source-visible music, singing, radio, or instrument playing
+in the main description. Use `N/A` when no score is wanted.
+
+## Timeline notation
+
+- Start the opening with `[Shot 1]` and no timestamp.
+- Mark each later cut with a sequential label and a strictly increasing time,
+  for example `[Shot 2] At 00:03.500, ...`.
+- Use a new shot for an actual edit or meaningful viewpoint/time change. Keep a
+  continuous action or morph in one shot and describe its internal timing.
+- Describe camera motion as an action: type, direction, useful amplitude, and
+  speed. Do not append a disconnected list of camera keywords.
+- Keep every cut and event within the requested clip duration.
+
+## Speech, text, and sound
+
+Give each speaking or singing source a stable ID such as `(S1)` or `(S2)` and
+reuse it across shots. Put only the language tag and literal words inside the
+dialogue wrapper:
+
+```text
+The courier (S1) says softly: <d>[English] The gate is open.</d>
+```
+
+Preserve supplied dialogue, lyrics, punctuation, and visible text verbatim. State
+when a voice is off-screen and keep the visible character's lips closed when that
+is required. If a line spans a cut, describe the audio continuity on both sides
+of the cut. Place visible signs, labels, and interface copy in double quotes.
+
+Tie diegetic sounds to the event that produces them. Avoid describing the same
+sound as both diegetic and non-diegetic.
+
+## Anchor-specific checks
+
+- I2VA: begin from the supplied frame rather than merely borrowing its style.
+- FL2VA: describe intermediate motion; do not provide two unrelated still-image
+  descriptions. Reach Picture 2 at the actual end time.
+- L2VA: keep the final picture assigned to the last shot, not automatically to
+  Shot 1, and make the preceding action physically plausible.
+- All modes: reference labels must map to supplied assets and remain stable.
+
+Before delivery, verify the field order, shot numbering, timestamps, reference
+labels, dialogue fidelity, requested language, duration, and absence of unresolved
+template variables.

.agents/skills/h3-prompt-writing/references/portable-workflow.md

@@ -0,0 +1,121 @@
+# Portable H3 production workflow
+
+This is the runtime contract for the nine H3 skills in this repository. Read it
+alongside the selected creative skill. The bundle uses local files and available
+tools; no proprietary canvas, choice-card widget, planner, or executor is required.
+
+## Scope and decisions
+
+- Resolve the requested deliverable first: prompt, storyboard, assets, or finished
+  video. A prompt-only request ends with the prompt; it does not start generation.
+- Reuse the user's duration, language, style, model, audio policy, supplied assets,
+  and existing authorization. State reasonable defaults for optional gaps. Ask
+  only about missing inputs or choices that materially affect the requested work.
+- Creative checkpoints are reviewable artifacts, not mandatory repeated approvals.
+  Honor an explicit request to approve a preview before rendering. When the user
+  has authorized the complete workflow or delegated creative choices, show the
+  plan and continue within that scope without requiring a special UI control.
+- A skill never grants permission to publish, send messages, spend outside an
+  authorized budget, or change models against the user's choice. User constraints
+  override a preset's style, language, shot count, timing, or audio defaults.
+
+## Tools and durable artifacts
+
+1. Inspect the actual available tools, existing scripts, endpoint documentation,
+   and supplied media. Do not invent a tool name, model parameter, or returned path.
+2. Write plans and prompts to the user-selected project folder. If none is given,
+   use a new named folder under `outputs/`, outside the skill package. Keep useful
+   artifacts such as `brief.md`, `storyboard.md`, `prompts/`, `references/`,
+   `clips/`, and `final/`; create only those needed for this deliverable.
+3. Record each reference's role and path: identity, environment, style, typography,
+   first/last frame, or audio. Tie generated clips to the prompt revision, input
+   assets, model, settings, and output paths. Update downstream references when an
+   asset changes; retain previous outputs instead of overwriting user files.
+4. Use available image generation/editing tools for requested raster assets. Use
+   an available video endpoint or local service for clips, and available audio
+   tooling for requested standalone speech/music. For assembly and inspection,
+   use an installed editor or tools such as FFmpeg, ffprobe, and OpenCV. Follow the
+   host's tool policies, including its preferred image generation mechanism.
+5. Independent tool jobs may run concurrently when resources permit; this skill
+   does not require separate agents or delegated generation.
+6. If a required capability is missing, finish the useful in-scope preparation:
+   complete prompts, asset-role manifest, timeline, and verified request example
+   where possible. State the missing capability and unfinished media step. Do not
+   silently substitute another model, claim a render happened, or call a written
+   plan a completed video. Resume generation when the capability becomes available.
+
+## Compile the creative plan into H3 prompts
+
+Read [base-format.md](base-format.md) for T2VA / I2VA / FL2VA / L2VA, or
+[ref-format.md](ref-format.md) for Ref2VA. These are writing modes, not guaranteed API
+task names. Check the selected backend's input mapping before sending a request.
+
+- Base prompts use `integrated_multimodal_description`, `overall_soundscape`,
+  `non_diegetic_music`, in that order, with the appropriate keyframe preamble.
+- Ref2VA uses `subject_definitions`, `summary`, `retention_analysis`,
+  `detailed_description`, `overall_soundscape`, `non_diegetic_music`, in that order.
+- Every reference label must resolve to a supplied asset, in the same order used
+  by the backend. Omit nonexistent optional references and their dependencies.
+- Keep the creative brief in the user's language. Structured H3 rendering prompts
+  use English prose and exact field labels, while dialogue, lyrics, and visible
+  text retain their requested language and wording. If the user explicitly wants
+  only a same-language creative prompt, deliver that without silently replacing it
+  with English. Compile a separate rendering prompt when generation is requested.
+- `[Shot 1]` has no initial timestamp. Later shot labels mark actual cuts. For a
+  continuous morph, describe timed actions inside one shot rather than turning
+  each form into a new shot. Express camera direction, speed, and useful amplitude.
+- Separate image style from character identity. A style reference must not inject
+  its people, scenery, layout, typography, texture, or lighting into unrelated
+  shots. For literal text fidelity, inspect the rendered result; a prompt is not
+  proof that the exact lettering or beat synchronization was achieved.
+
+## Longer videos and audio
+
+Use the actual backend's supported duration, dimensions, frame grid, reference
+count, audio modes, and model partition. Inspect this checkout's H3 implementation
+or serving documentation before building requests. Do not assume that a feature
+branch's long-video extension exists on main or on a hosted service.
+
+For a verified vLLM-Omni Ref2VA continuation backend, `long_video=true` with
+`long_video_mode=continuation` can accept one `continuation_prompts` entry per
+planned window only when the Ref2VA request is executed with its local text
+encoder. Step execution and stages that use an external text encoder reject
+`continuation_prompts`; omit the field and use the supported per-step inputs in
+those configurations. Compute the window count from the aligned total frames,
+window size, and overlap. Each prompt uses local window time; track global
+shot/music time separately. Repeated overlap is guidance, not additional output
+duration. Do not use this mode on a task/partition that does not support it.
+
+Otherwise split into supported clips and assemble them. Same-scene clips may use
+tail/head keyframe continuity when supported. Scene changes should have explicit
+new scene references; a previous scene's latent tail can preserve the wrong
+setting. Choose cuts, morphing, and transitions from the brief rather than imposing
+hard cuts whenever a work exceeds 15 seconds. Generation windows and visual shots
+are different: one window may contain cuts, and one shot may span multiple windows.
+
+For supplied music, use the selected excerpt as the master timeline unless the
+user requests replacement. For native audio, write the musical/SFX direction in
+the prompt and verify that the backend actually returns audio. Do not invent a
+generic `generate_audio` parameter. If available, locked-source audio latents guide
+generation but the audio-VAE reconstruction is not bit-identical to the source.
+Use explicit final audio stream-copy when an exact original track is required.
+Prompted BPM is a target, not measured timing. Avoid automatically adding lyrics,
+voiceover, subtitles, or multiple competing scores to an instrumental request.
+
+## Verification and bounded recovery
+
+Inspect reference images before using them. After rendering, inspect representative
+frames, transitions, opening/ending states, text, identities, and scene continuity.
+Check audio by listening when possible; report any limit on audio review. Check
+duration, dimensions, fps, frame count, audio presence, and decode validity with
+media tools. If preserving audio exactly, compare extracted audio stream hashes.
+
+Correct a demonstrated failure with a targeted change to the affected prompt,
+reference, or edit. Keep raw generation alongside edited/upscaled versions and
+label the difference. Use an agreed retry/resource budget; absent one, allow one
+targeted rerender, then report persistent defects and concrete options rather than
+entering an unbounded generation loop. Do not switch providers automatically.
+
+Deliver actual artifact paths, the prompts/settings needed to repeat the work,
+and material visual/audio limitations. Distinguish requested behavior, observed
+model output, and post-production repairs.

.agents/skills/h3-prompt-writing/references/ref-format.md

@@ -0,0 +1,105 @@
+# H3 full-reference prompt format
+
+Use this reference for Ref2VA prompt writing. It is a compact,
+repository-authored description of the output contract, not a copy of an
+upstream prompt guide.
+
+## Emit six fields in order
+
+```text
+subject_definitions: ...
+
+summary: ...
+
+retention_analysis: ...
+
+detailed_description: [Shot 1] ...
+
+overall_soundscape: ...
+
+non_diegetic_music: ...
+```
+
+Write the structural prose in English. Preserve the requested language and exact
+wording for dialogue, lyrics, and visible text.
+
+## Reference labels
+
+Assign labels only to supplied content and keep every label stable across all six
+fields:
+
+- `<Subject N>` identifies reusable visible content such as a person, object,
+  environment, costume, interface, action, or visual treatment.
+- `<Picture N>` identifies an image used as a concrete frame, composition, or
+  storyboard anchor. If an image only defines a subject, cite it in that subject's
+  definition instead of inventing a separate target-frame role.
+- `<Video N>` identifies a source video being edited or continued, or a video
+  whose temporal structure is referenced.
+- `<Audio N>` identifies a supplied audio signal that is copied or referenced.
+  Audio and video label numbers are independent even when they came from one file.
+
+In `subject_definitions`, give each tracked item a separate line. State its source,
+role, and the features that later sections must preserve or transform. Do not
+create labels for assets or roles that do not exist.
+
+## Summary and retention
+
+`summary` is one short paragraph describing the target and the principal reference
+relationships. Use only labels already defined. Distinguish these operations:
+
+- a concrete image frame anchor;
+- generation guided by identity, style, action, camera, or audio characteristics;
+- direct editing of a source video;
+- continuation from a source video;
+- copying an audio signal;
+- referencing audio properties without copying the signal.
+
+`retention_analysis` has one line per label and states where it is used and what
+changes. Use consistent machine-readable relationship values:
+
+- Visual: `fully_preserved`, `partially_preserved`, `attribute_transfer`, or
+  `weak_reference`.
+- Audio: `fully_copy`, `partially_copy`, `reference`, or `weak_reference`.
+
+Explain the concrete retained and changed traits after the relationship value.
+Do not claim full preservation when identity, geometry, timing, or signal content
+is intentionally altered. Do not count newly requested target action as an
+unintended loss of reference fidelity.
+
+## Detailed description
+
+`detailed_description` is the playback-ordered rendering plan. For each shot,
+cover the current composition, subject appearance and position, environment,
+lighting, actions and state changes, camera behavior, diegetic sound, and the
+point at which each reference becomes relevant.
+
+- Start with `[Shot 1]` and no timestamp.
+- Give later cuts sequential labels with strictly increasing times, such as
+  `[Shot 2] At 00:03.500, ...`.
+- Place relevant reference labels next to the described subject, frame, motion,
+  or sound; do not collect detached labels at the end.
+- Use stable `(S1)`, `(S2)`, and subsequent IDs for vocal sources. Put literal
+  speech or lyrics inside `<d>[Language] ...</d>`.
+- Preserve supplied visible text verbatim and place it in double quotes.
+- Describe continuous transformations inside one shot unless an actual cut is
+  intended. State camera direction, speed, and amplitude when useful.
+
+For video editing, explicitly identify what stays from the source timeline and
+what is replaced. For continuation, distinguish the inherited opening state from
+new content. For style reference, transfer visual properties without accidentally
+copying people, scenery, layout, or typography.
+
+## Audio fields
+
+`overall_soundscape` summarizes ambience, physical action sounds, and non-verbal
+human sounds. `non_diegetic_music` describes score audible only to the audience.
+Keep dialogue and diegetic music in `detailed_description`. Use `N/A` only when
+the corresponding layer is intentionally absent.
+
+If audio is copied, state whether the whole signal or only a range/layer is used.
+If it is referenced, describe the specific property—such as voice timbre, tempo,
+rhythm, or texture—without claiming signal reuse.
+
+Before delivery, verify the six-field order, one-to-one label definitions,
+retention coverage, shot timing, dialogue fidelity, requested language, and
+absence of unresolved template variables.

.agents/skills/handdrawn-live-video-generator/SKILL.md

@@ -0,0 +1,69 @@
+---
+name: handdrawn-live-video-generator
+description: Write or generate a continuous live-action scene with a rough luminous hand-drawn entity that contacts real objects, morphs, and escapes a slightly delayed handheld camera. Use for gentle surreal mixed-media shorts, not flat-vector-only animation or horror clips.
+metadata:
+  compatibility: Portable prompt writing with sibling h3-prompt-writing; actual generation requires an available video backend and any referenced media.
+---
+
+# Handdrawn Live-Action Fusion
+
+Read the [portable workflow](../h3-prompt-writing/references/portable-workflow.md).
+Keep the user's requested scope and language. A creative-prompt request delivers
+the prompt, without generating media. For a requested finished video, prepare
+the prompt and proceed under existing authorization using available tools.
+
+## Creative contract
+
+The default is a 15-second, 16:9 continuous scene in an everyday real space. A
+flat luminous drawn entity visibly contacts a real hand/object early, transforms
+as one traceable being, escapes through connected space, and the handheld camera
+reacts slightly late. Duration and ratio follow the user's explicit choices.
+
+Use rough crayon, chalk, colored pencil, pastel, or brush strokes with uneven fill,
+frayed edges, line jitter, and frame-by-frame redraw. Keep the live-action setting
+physically grounded and the drawn presence visibly planar. Uniform neon tubes,
+plush characters, polished CG, and clean vector strokes would change this style.
+
+Choose a fresh setting, entity, palette, contact mechanism, transformation chain,
+escape route, camera reaction, and emotional ending within the user's constraints.
+Do not ban an otherwise requested motif simply because an upstream example used
+it. The tone is warm, playful, and gently surprising rather than threatening.
+
+## Timed action structure
+
+For a 15-second brief, use these intervals as a starting structure:
+
+- 0–3s: establish obvious real/drawn contact, such as landing on a palm or slipping
+  around a finger. The real participant reacts.
+- 3–6s: begin a continuous morph while retaining a color smear, line, curve, or tail
+  from the previous shape. Give it a playful action.
+- 6–10s: let it escape into an adjacent part of the same space. The camera pans,
+  tilts, or moves only after the entity starts leaving its view.
+- 10–13s: add another readable interaction or surprise, keeping the same entity.
+- 13–15s: extend an earlier drawn motif across the space into a gentle larger
+  transformation, then resolve with a small emotional/comic detail.
+
+Retiming changes the intervals proportionally or reorganizes beats; do not leave
+15-second timestamps in a shorter prompt. Preserve a user's explicit final form
+instead of forcing the spatial finale when it conflicts with their ending.
+
+## Prompt delivery and rendering
+
+For a same-language creative prompt, begin with duration, ratio, real space, and
+the live-action/drawing fusion. Then describe phone-camera texture, the timed
+action intervals, drawn material, camera lag, specific exclusions, and ambient
+sound. Avoid unrelated headings or an essay when the user asked for copy-ready text.
+
+For actual H3 generation, create a separate structured rendering prompt using the
+appropriate base/Ref2VA guide: English prose with literal dialogue/visible text in
+the requested language. Preserve the creative intent and actual media bindings.
+Do not send unsupported model settings or promise exact first-frame conditioning
+merely because an image is present.
+
+## Review
+
+Check clear early physical contact, one continuous entity, connected geography,
+retained shape motifs, delayed camera reaction, rough planar drawing, and a
+non-threatening ending. Avoid jump scares, sudden blackouts, menacing anatomy, and
+unrequested extra characters. Deliver the actual prompt/video and state any
+remaining contact, continuity, texture, or camera-timing defects.

.agents/skills/handdrawn-live-video-generator/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "Handdrawn Live-Action Fusion"
+  short_description: "Direct drawn morphs and real-world interactions"
+  default_prompt: "Use $handdrawn-live-video-generator to prepare the requested H3 prompts or video with the available tools."

.agents/skills/minimalist-product-ad-generator/SKILL.md

@@ -0,0 +1,71 @@
+---
+name: minimalist-product-ad-generator
+description: Plan, prompt, or produce minimalist product advertising shorts from product images, preserving variants, material, color, concise typography, and rhythmic camera motion. Use for product launches and ecommerce films, not presenter-led ads or complex UI demonstrations.
+metadata:
+  compatibility: File-based agent workflow. Requires the sibling h3-prompt-writing package; rendering uses available image, video, audio, and editing tools.
+---
+
+# Minimalist Product Ad Generator
+
+Read the [portable workflow](../h3-prompt-writing/references/portable-workflow.md).
+Use [product direction](references/product-direction.md) when planning anchors,
+copy, and timed beats. Respect the requested output: prompt-only work needs no
+generation tool, while a finished film includes generation, inspection, and assembly.
+
+## Resolve the product brief
+
+Inspect supplied product photos, packaging, renders, and detail shots. Record the
+product category, main variant, actual body/accent colors, material finish, usable
+mechanisms, and requested copy. Preserve identity and product facts; visual polish
+does not authorize recoloring the product or inventing a feature.
+
+Reuse the specified duration, ratio, language, and model. With no constraints,
+suggest a 10-second H3 film and one of white studio, dark rim light, brand color
+field, or light lifestyle staging. Ask for a missing product reference when exact
+identity is required; distinguish a speculative concept from an existing product.
+If the user delegates direction, record the adopted choices and proceed.
+
+## Design the film
+
+1. Choose a product-specific spine: launch reveal, mechanism interaction, or color
+   family. Describe concrete actions such as a lid opening or a crown turning.
+2. Write short sensory/benefit copy grounded in supplied facts. Preserve the user's
+   exact approved wording and language. For newly requested English copy, 3–5 words
+   usually fit one readable line; do not force English on other-language campaigns.
+3. Plan three independent anchors: hero silhouette, material/mechanism detail, and
+   final product/copy composition. Generate only the assets needed or requested.
+   Do not combine them into a panel sheet that the video model might reproduce.
+4. Keep product color, light direction, ratio, and finish consistent across anchors.
+   Original photos establish identity; generated anchors must pass that identity
+   check before they can replace them as rendering references.
+5. Build the beat table from the reference. Include setup, action, impact, and a
+   stable closing composition; vary shot distance only when it reveals something.
+
+## Prompt and render
+
+Compile each beat into the appropriate H3 fields using the sibling writing guide.
+Bind every anchor to its actual role. Include exact visible copy, entry/exit times,
+colors, and same-line text motion in the rendering prompt, not only in the plan.
+Keep typography in intentional negative space without covering the product. A
+copy-reference image does not mean its line must persist throughout the whole film.
+
+For a short ad, prefer one coherent full-frame film when the backend supports it.
+Keep variants ordered: main variant first, supporting variants later, full family
+only when useful at the close. Avoid product walls, grids, or an anchor-sheet ending.
+
+Use native audio when appropriate to the brief. A restrained electronic score may
+use crisp plucks, airy pads, kick/sub accents and tactile wooden percussion. Set
+tempo and energy from the user's direction. Generate a separate score only if
+requested or needed for a disclosed audio repair, using an available tool.
+
+## Review and deliver
+
+Verify product geometry, variant colors, functional action, final copy, aspect
+ratio, and closing readability. Listen for excessive bass, dropouts, and unfinished
+endings. If editing to supplied music, select a suitable continuous excerpt and
+align visual impacts to observed beats; do not claim synchronization from text alone.
+
+For bad typography, first reduce simultaneous text or correct its reference. Exact
+lettering may need a disclosed overlay when consistent with the user's goal; keep
+the raw output. Deliver the film, prompt/beat plan, used copy and music excerpt,
+and any remaining product/text fidelity issues.

.agents/skills/minimalist-product-ad-generator/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "Minimalist Product Ads"
+  short_description: "Direct product anchors, copy, motion, and audio"
+  default_prompt: "Use $minimalist-product-ad-generator to prepare the requested H3 prompts or video with the available tools."

.agents/skills/minimalist-product-ad-generator/references/product-direction.md

@@ -0,0 +1,52 @@
+# Product anchors, typography, and beats
+
+## Narrative choices
+
+- Launch: negative space → hero product → material detail → real mechanism action
+  → supporting variants → stable product and copy.
+- Feature touch: stillness → interaction trigger → visible mechanism → detail
+  magnification → useful result → closing.
+- Color family: main variant alone → supporting colors enter → ordered relation
+  → unified material finish → full set and copy.
+
+## Independent anchor briefs
+
+| Anchor | What it locks | Check |
+| --- | --- | --- |
+| Hero | Main variant, recognizable silhouette, attractive opening view | Match supplied body color and product structure |
+| Detail | Material, reflection, texture, visible mechanism | No invented ports, controls, or functionality |
+| Closing | Product position, negative space, typography treatment | Real copy, readable contrast, no panel/grid layout |
+
+Each is a standalone full-frame image with the project ratio. Reference roles
+protect appearance and layout; they do not replace the storyboard or require three
+sequential static-image segments. Simplify the number of anchors if the backend
+cannot bind them all, and update the prompt labels accordingly.
+
+## Typography
+
+Use one readable line at a time, with a typeface appropriate to the brand. A clean
+semibold sans-serif is a useful default, not a requirement to copy a specific brand.
+For a two-part color treatment, use dark text on a light field or light text on a
+dark field, then emphasize the second part with a specific real product color.
+Check contrast rather than blindly coloring text like the product.
+
+The first phrase may enter gently; the second follows while the first shifts
+slightly within the same line. Do not accidentally turn this into stacked lines.
+Use actual copy in the prompt and spell out its timing. Avoid arrows, separators,
+or placeholder marks that the model may reproduce. For a 10-second ad, one middle
+copy moment and one stable closing line are often enough. Omit copy when requested.
+
+## Beat table
+
+| Time range | Shot purpose | Visual lead and camera | Variant state | Exact copy | Text color/motion | Handoff | Rhythm |
+| --- | --- | --- | --- | --- | --- | --- | --- |
+
+Use one main action per beat. For 5/10/15-second films, roughly 3–4/5–7/6–9 beats
+are planning suggestions, not required cuts. Record the ending product position
+and motion vector so the next beat can inherit them. Ensure the closing line has
+time to be read rather than using every second for motion.
+
+If replacing music, keep useful SFX unless replacement of the whole track was
+requested. Use a continuous excerpt with a clear entry, stable middle, and suitable
+ending. A short tail fade may resolve a click; do not hide an incomplete phrase
+behind a long fade. Confirm the entire result remains audible.

.agents/skills/music-video-subtitle-generator/SKILL.md

@@ -0,0 +1,74 @@
+---
+name: music-video-subtitle-generator
+description: Plan, write, audit, or produce music videos with coordinated music, imagery, camera motion, optional performance, and lyric typography. Use for beat-aware MVs and instrumental visual narratives, not ordinary subtitle cleanup or unrelated video edits.
+metadata:
+  compatibility: File-based workflow with sibling h3-prompt-writing; audio analysis, generation, and final assembly use available tools and backends.
+---
+
+# Music Video and Lyric Typography
+
+Read the [portable workflow](../h3-prompt-writing/references/portable-workflow.md)
+and, when building a timeline, [music timeline](references/music-timeline.md).
+Deliver the requested scope: audit, prompt, storyboard, or complete video.
+
+## Lock the musical and visual intent
+
+Identify duration, ratio, output resolution, music source/excerpt, visual style,
+character/scene references, performance mode, and whether text is wanted. Treat a
+supplied song as the master track unless the user requests replacement. Preserve
+provided lyrics exactly; do not translate or rewrite them unasked. For instrumental
+work, do not invent singing, lyrics, a performer, or typography. Generate original
+lyrics only when requested or clearly within an authorized songwriting brief.
+
+Measure supplied audio duration and inspect/listen for beat, phrase, breath, drop,
+and energy changes using available tools. Label estimated or planned timing when
+measurement is unavailable. If the requested excerpt exceeds the audio, resolve
+the discrepancy without silently stretching or looping it.
+
+## Reference roles
+
+- Character references control identity, silhouette, wardrobe, and presence.
+- Scene references control environment, time of day, spatial anchors, and lighting.
+- Typography references control letter treatment, layout, and motion, not people
+  or scenery visible in those references.
+- Music references control the requested track or sound style; distinguish copying
+  a supplied recording from generating similar instrumentation.
+
+Do not apply a preset's grain, darkness, hard cuts, or constant close-ups to an
+incompatible brief. A bright nighttime scene remains night. A continuous flat
+vector morph keeps smooth color fields and uses elastic shape motion for accents.
+Use hard cuts, slow holds, or dissolves only when appropriate to the user's intent.
+
+## Compose the timeline and H3 prompts
+
+Map visual actions, camera moves, scene changes, and optional text to the music's
+global time. Build a causal visual progression instead of defaulting to one person
+singing throughout. Allow room to read a shape, gesture, or lyric before changing it.
+
+For longer work, choose a supported continuation or clip-assembly approach through
+the portable workflow. Keep visual shot boundaries distinct from generation
+windows. Local window prompts must account for overlap without replaying the full
+previous action. A new scene may require a new reference rather than inheritance
+of the old scene. Record entry/exit states for each planned segment.
+
+Compile the timeline into H3's base or Ref2VA fields, preserving dialogue, lyrics,
+and visible text verbatim. Include concrete instrumentation and rhythmic accents,
+but treat requested BPM and beat alignment as targets until checked in the output.
+
+## Typography and finishing
+
+When requested, give each shot one main readable text event. Keep words away from
+eyes and critical lip motion. Match sung text to the active phrase, and distinguish
+spatial typography from subtitle bars. Reduce text complexity if the model cannot
+render it, or use a disclosed overlay consistent with the requested deliverable.
+
+For supplied audio, assemble clips to one global master track, maintaining phrase
+and lip continuity; avoid cuts inside a vowel. For native audio, inspect actual
+continuity across generation windows. Preserve useful SFX and avoid layered duplicate
+music. Retiming, grain, grading, interpolation, and upscaling are deliberate editing
+choices, not automatic finishing requirements.
+
+Verify transformation/story order, identities, scene changes, requested text,
+musical continuity, timing, and full media decode. Label raw and processed versions.
+Deliver the final file and current prompt/timeline files, with measured properties
+and remaining limitations rather than claims of perfect synchronization.

.agents/skills/music-video-subtitle-generator/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "Music Video Direction"
+  short_description: "Plan music-driven scenes and optional lyric text"
+  default_prompt: "Use $music-video-subtitle-generator to prepare the requested H3 prompts or video with the available tools."

.agents/skills/music-video-subtitle-generator/references/music-timeline.md

@@ -0,0 +1,40 @@
+# Global music and local video timing
+
+Use this planning skeleton, omitting vocal/text fields when not requested:
+
+```text
+Global: music/excerpt, duration, ratio, character/style/scene locks, cut policy.
+
+Segment S01 — global start/end; local generation start/end
+Incoming state:
+Music cue: measured beat/phrase timestamp, or clearly labeled planned accent
+Vocal line and speaker (if present):
+Visible typography (if requested):
+Visual action and camera:
+Outgoing state and transition:
+References with exact roles:
+```
+
+Keep one global music clock. For measured tempo B, the nominal quarter-note spacing
+is 60/B seconds, but real tracks may have pickups, swing, tempo changes, or inaccurate
+automatic detections. Validate important cut points against the waveform/listening.
+An onset detector also finds SFX and melodic attacks; it does not prove drum beats.
+
+Choose a phrase-complete excerpt when possible. Track breaths, lyric pauses, and
+vowel continuity for a performance MV. For an instrumental morph, accent the
+physical preparation, transformation impact, and release rather than introducing
+unrequested cuts. The final shape or story payoff needs a readable ending.
+
+For continuation, maintain a table of absolute output frames and local window
+times. Use the verified backend's native alignment and overlap behavior. Opening
+overlap continues the inherited state; new actions belong after it. Do not sum
+overlap twice or describe every window as a fresh opening shot.
+
+For separately generated clips, use a previous tail frame only when the next shot
+continues the same scene and the backend supports that keyframe role. For an
+intentional cut, retain appropriate identity references and select the new scene.
+Preserve screen direction or a useful match element when it helps the transition.
+
+When text is requested, track exact string, language, entry/exit time, screen
+region, and motion. Typography style cards must not donate their depicted characters
+or settings. Preserve user text; do not silently shorten lyrics to fit a template.

.agents/skills/paper-collage-explainer-generator/SKILL.md

@@ -0,0 +1,69 @@
+---
+name: paper-collage-explainer-generator
+description: Translate narration, ideas, or knowledge points into editorial halftone paper-collage metaphors, still prompts, storyboards, or stop-motion clips. Use for photographic cutout collage and tactile assembly, distinct from paper-puppet dioramas or generic cartoons.
+metadata:
+  compatibility: File-based workflow with sibling h3-prompt-writing; media creation uses available image/video tools and an editor when needed.
+---
+
+# Paper Collage Explainer Generator
+
+Read the [portable workflow](../h3-prompt-writing/references/portable-workflow.md)
+and [collage design](references/collage-design.md) when constructing scenes.
+Honor prompt-only and asset-only requests as well as finished video requests.
+
+## Meaning and media policy
+
+Extract each source line's meaning, emotion, action verb, visual metaphor, and
+three to six readable object groups. Preserve the user's topic and wording as
+context; do not put narration on screen merely because it is available.
+
+Default to tactile paper SFX, with no BGM, voiceover, or subtitles unless requested.
+An explainer does not inherently require speech. Existing user requests for those
+layers already establish the choice; do not ask again. A short 16:9 clip of roughly
+four seconds is a useful default for one assembly beat, subject to backend limits.
+
+## Plan the image and motion together
+
+Create a concise plan of metaphor, palette, object groups, assembly order, duration,
+and sound cues. For a series, keep paper texture, halftone treatment, keylines,
+shadows, and rhythmic language consistent; vary the metaphor and color field only
+where the meaning benefits. A user's no-text or no-music requirement applies to
+every downstream asset, prompt, clip, and edit.
+
+Design each still as the completed final composition. Use black-and-white
+photographic halftone cutouts, selective colored cardstock, warm cream keylines,
+subtle fiber/torn edges, and soft physical layer shadows on a bold color field.
+Keep layered depth and negative space; avoid excessive age, dirt, wrinkles, or
+automatic brown/kraft backgrounds. Exact editable layers require a compositing
+workflow, not a claim that a generated raster video is a layered project.
+
+Generate final-composition stills when requested/needed and inspect them. Follow
+any requested approval gate; otherwise continue within the authorized workflow.
+
+## Animate the assembly
+
+Start from a clean paper field matching the planned final color. Base pieces enter,
+then the main metaphor, then supporting objects. Each piece slides or pops in,
+bounces slightly, presses flat, pauses, and locks. End with a brief hold on the
+completed composition. Use stop-motion assembly rather than global fades, smooth
+digital drifting, fast spinning, or liquid morphs unless explicitly requested.
+
+Use the supplied final still as a last-frame anchor only when the backend supports
+it. Otherwise identify it honestly as composition/style guidance; do not describe
+a reference image as a guaranteed final-frame constraint. Compile the actual H3
+mode with the sibling guide and bind only real inputs.
+
+Map short paper slides, pops, taps, rustles, and snaps to motion. Preserve those
+SFX during assembly. When the user adds music or narration later, mix with the
+useful SFX unless they request full replacement.
+
+## Review and deliver
+
+Check the metaphor without relying on text, object count/readability, matching
+opening color, visible piece-by-piece construction, final-frame resemblance,
+controlled paper texture, and no unwanted UI/letters/music/voiceover. If the
+assembly is weak, simplify the groups and make their arrival order concrete.
+
+For multiple clips, preserve the planned order and SFX. Deliver prompts, current
+stills, and actual clips/assembly as requested. Explain any missing generation
+capability or remaining drift; keep raw output separate from edited fixes.

.agents/skills/paper-collage-explainer-generator/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "Paper Collage Explainers"
+  short_description: "Animate editorial paper metaphors with tactile SFX"
+  default_prompt: "Use $paper-collage-explainer-generator to prepare the requested H3 prompts or video with the available tools."

.agents/skills/paper-collage-explainer-generator/references/collage-design.md

@@ -0,0 +1,33 @@
+# Editorial collage design
+
+For each source idea record meaning, emotion, one active verb, a concrete visual
+metaphor, three to six object groups, and a tactile sound cue. For example, a
+collaboration concept can become separate cutout hands fitting paper routes
+together; it need not become a screen full of the word "collaboration".
+
+Style signature: flat bold color field, black-and-white halftone photographic
+cutouts, selective colored cardstock accents, warm cream keylines, soft physical
+paper shadows, fine uncoated fiber, refined irregular torn edges, layered seams.
+Control texture so the video feels handmade without becoming dirty or distressed.
+
+Palette choices may support meaning: orange/red for urgency, mustard for tools or
+warning, green for reflection, purple for memory, teal for collaboration, rose for
+theatrical contrast. These are suggestions, not universal semantic rules. Use the
+user's palette and avoid an unrelated kraft-paper opening.
+
+Still specification: source meaning → metaphor → ratio → background/accent colors
+→ object roles → foreground/midground/background → final composition → exclusions.
+
+Motion specification: matching empty color field → base paper structures → main
+subject → secondary groups → light bounce/press-flat/pause → completed composition.
+Explicitly describe material boundaries and which piece moves at each beat. A
+static camera is usually enough to reveal assembly; add movement only when useful.
+
+Sound specification: synchronize restrained slides, pops, taps, rustles, and snaps.
+Set `non_diegetic_music: N/A` for the default SFX-only H3 prompt. Do not use silence
+as a synonym for no background music. Dialogue and subtitles are optional layers.
+
+Review one established clip as a style/cadence reference for subsequent clips.
+Change the specific failure: fewer groups for clutter, more physical arrival
+steps for digital drifting, less aging for dirty paper, stronger final state for
+composition drift. Do not apply an unrelated aesthetic preset to fix one defect.

.agents/skills/papercraft-stop-motion-explainer/SKILL.md

@@ -0,0 +1,68 @@
+---
+name: papercraft-stop-motion-explainer
+description: Turn knowledge topics into layered papercraft stop-motion explainers, including visual metaphors, miniature sets, prompts, storyboards, motion and sound plans, or rendered video. Use for tactile paper dioramas and mechanisms, not ordinary vector cartoons or photographic collage.
+metadata:
+  compatibility: Portable file-based workflow with sibling h3-prompt-writing; actual media generation uses tools available in the host environment.
+---
+
+# Papercraft Stop-Motion Explainer
+
+Read the [portable workflow](../h3-prompt-writing/references/portable-workflow.md).
+For scene/prompt planning read [paper staging](references/paper-staging.md).
+Route a single image prompt, image series, short animation prompt, or storyboard
+request directly to that artifact; do not force a full film workflow.
+
+## Knowledge and direction
+
+Identify the exact topic, source facts, audience, desired learning outcome, and
+output type. Preserve technical terms and causal meaning. A visual metaphor helps
+explain a mechanism; it must not turn into a false scientific claim.
+
+Reuse requested duration and ratio. If unspecified, a 30-second, 16:9 general-audience
+concept is a reasonable starting point. Offer distinct directions only when useful:
+pop-up-book journey, paper laboratory, layered cross-section, miniature nature
+theater, or mechanism board. Each should identify one learning outcome, a paper
+mechanism that conveys it, and the signature visual moment.
+
+## Build the paper world
+
+Design characters only if they help the explanation. Specify paper construction,
+silhouette, joint/hinge design, replaceable facial parts, gestures, and repeatable
+identity cues. Keep the central knowledge object readable.
+
+Plan foreground occluders, midground mechanism, background scenery, and a distant
+paper plane when the shot benefits from depth. Record each layer's material,
+thickness, cut/fold edges, shadow, movement, and educational role. Use real paper
+mechanics such as pull-tabs, sliders, rotating discs, page folds, and pinned joints.
+Motion is stepped with small pauses and rebounds, not a smooth CG melt.
+
+Create an asset list by function: characters, scientific objects, information cards,
+environment, and moving mechanisms. For a complete production, prepare one to three
+preview briefs or images when useful and authorized. Respect any preview approval
+requested by the user; otherwise continue under the existing end-to-end scope.
+
+## Prompts and timeline
+
+Write a compact storyboard with columns: time, knowledge beat, visual action,
+paper movement, camera/transition, sound cue. Each shot explains one knowledge
+step, with a readable pause for a necessary label. Keep ratio, material, light
+direction, identity, and domain vocabulary consistent across image-series prompts.
+
+For a short image-to-video prompt, begin from the actual image and animate a few
+paper mechanisms with restrained push-in or lateral parallax. Preserve thickness,
+cut edges, texture, and layer shadows. Compile final rendering prompts using the
+bundled H3 mode guide, with truthful references and local clip timing.
+
+## Sound, generation, and review
+
+Map paper flips, cardboard slides, joint taps, soft pops, and rustles to actual
+motion. Match any requested music to the topic and cultural context rather than
+automatically using the same whimsical score. Keep speech intelligible. If narration
+is requested, preserve its approved wording, measure its duration, and adjust the
+timeline or script within the user's scope; do not silently speed up the voice.
+
+Use available generation/editing tools, then inspect both educational accuracy
+and paper construction: readable mechanism, separate layers/shadows, hinged or
+stepped motion, plausible camera movement, and a clear ending takeaway. Check
+narration tails, clicks, audio gaps, and music balance. Deliver requested artifacts
+and state missing rendering capabilities or unresolved scientific/visual issues.

.agents/skills/papercraft-stop-motion-explainer/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "Papercraft Explainers"
+  short_description: "Explain ideas with layered paper mechanisms"
+  default_prompt: "Use $papercraft-stop-motion-explainer to prepare the requested H3 prompts or video with the available tools."

.agents/skills/papercraft-stop-motion-explainer/references/paper-staging.md

@@ -0,0 +1,48 @@
+# Paper staging and prompt building blocks
+
+## Materials and depth
+
+Use layered matte cardboard, visible thickness, fibers, folds, seams, tabs, brads,
+cut edges, and physical shadows. Separate foreground, mechanism, background, and
+far background with spacing. Roughly 4–7 layers often suffice; remove layers that
+hide the explanation. Place reading-heavy labels on stable planes. Use miniature
+macro photography and moderate depth of field without blurring necessary facts.
+
+| Plane | Useful content | Motion | Purpose |
+| --- | --- | --- | --- |
+| Foreground | Paper leaves, frame, curtain, props | Small parallax or hinged reveal | Depth without concealing the mechanism |
+| Midground | Knowledge model, host, arrow/cards | Pull-tab, slider, joint, rotating disc | Main cause and effect |
+| Background | Diorama set or schematic layers | Slow paper rail shift | Context and hierarchy |
+| Far plane | Paper sky/board | Restrained change | Stable visual support |
+
+## Prompt ingredients
+
+Image: topic and core object → miniature diorama → layered paper construction →
+specific material edges → physical light/shadows → clear depth → necessary labels.
+One frame should communicate one main idea, not the entire lesson at once.
+
+Image series: repeat the style/identity anchors, then state what changes and why
+that state explains the next knowledge step. Keep the same ratio and light direction.
+
+Short animation: preserve the supplied image, make a small paper mechanism move,
+include tiny pauses/rebounds, and select one restrained camera move. Small puppet
+gestures and background rails can provide life without overwhelming the model.
+
+## Camera and transitions
+
+Use slow push-in, lateral parallax, static demonstration, macro detail, or a slight
+top-down cross-section view. A full orbit requires a set built for that view.
+Transitions may use a page flip, pop-up unfold, pull-tab, paper cloud, circular
+mask, cut-paper door, layer separation, confetti, or sticker reveal. Avoid liquid
+morphing, neon glitches, and glass shatter unless explicitly requested as contrast.
+
+## Sound and exclusions
+
+Use a few motion-specific accents: paper flip, scissors, cardboard slide, joint
+tap, tape peel, or box opening. A score, when appropriate, can use light percussion,
+plucked instruments, or topic-specific instrumentation. Do not infer narration
+merely because the video teaches something.
+
+Keep constraints concise and relevant: no glossy plastic, smooth CG morphs, metal
+surfaces replacing paper, missing thickness/shadows, or frantic camera movement.
+Include them in supported prompt fields; do not invent a negative-prompt API field.

.buildkite/amd/scripts/bootstrap-amd-omni.sh

@@ -60,7 +60,8 @@ resolve_test_specs() {
     if ! selected_specs=$(python3 "$AMD_SUITE_SELECTOR" \
         --branch "$BUILDKITE_BRANCH" \
         --labels "$PR_LABELS" \
-        --debug-test-yaml "${DEBUG_TEST_YAML:-}"); then
+        --debug-test-yaml "${DEBUG_TEST_YAML:-}" \
+        --nightly "$NIGHTLY"); then
         exit 1
     fi
 
@@ -80,6 +81,12 @@ filter_test_specs_by_skip_ci() {
         case "$suite_spec" in
             READY_TESTS:*) level="l2" ;;
             MERGE_TESTS:*) level="l3" ;;
+            NIGHTLY_TESTS:*)
+                # Explicit PR and scheduled nightly selections must survive
+                # L2/L3 diff gating, including docs-only main commits.
+                runnable_specs+=("$suite_spec")
+                continue
+                ;;
             *)
                 echo "ERROR: unknown AMD test suite spec '$suite_spec'" >&2
                 exit 1
@@ -101,7 +108,7 @@ filter_test_specs_by_skip_ci() {
                 ":memo: CI skipped — docs or pytest skip-mark changes only" \
                 --style "info" 2>/dev/null || true
         fi
-        echo "No AMD L2/L3 suites remain after skip-ci filtering."
+        echo "No AMD suites remain after skip-ci filtering."
         exit 0
     fi
 }
@@ -219,8 +226,8 @@ if [[ $BUILDKITE_BRANCH == "main" ]]; then
     file_diff=$(get_diff_main)
 fi
 
-# Resolve PR tier labels before skip-ci so ready and merge suites can be
-# filtered independently for CI-YAML-only changes.
+# Resolve PR tier labels before skip-ci so ready/merge suites can be filtered
+# independently while an explicitly selected nightly suite passes through.
 resolve_test_specs
 filter_test_specs_by_skip_ci
 

.buildkite/amd/scripts/build-ci-image.sh

@@ -0,0 +1,69 @@
+#!/usr/bin/env bash
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+
+set -euo pipefail
+
+# Build and push the ROCm CI image with a registry-backed BuildKit cache. The
+# AMD builders are ephemeral, so a local Docker cache cannot survive between
+# jobs. Keeping the cache beside the runtime image also reuses the registry
+# credentials already configured on the amd-cpu agents.
+
+REGISTRY="${ROCM_CI_IMAGE_REGISTRY:-rocm/vllm-omni}"
+DOCKERFILE="docker/Dockerfile.rocm"
+BUILDER_NAME="vllm-omni-rocm-builder"
+ROCM_ARCH="${PYTORCH_ROCM_ARCH:-gfx942}"
+
+if [[ -z "${BUILDKITE_COMMIT:-}" ]]; then
+    echo "ERROR: BUILDKITE_COMMIT is not set" >&2
+    exit 1
+fi
+
+dependency_files=(
+    "$DOCKERFILE"
+    "${DOCKERFILE}.dockerignore"
+    pyproject.toml
+    setup.py
+    tools/install_torchcodec_rocm.sh
+)
+while IFS= read -r dependency_file; do
+    dependency_files+=("$dependency_file")
+done < <(find requirements -maxdepth 1 -type f -name '*.txt' -print | LC_ALL=C sort)
+
+for dependency_file in "${dependency_files[@]}"; do
+    if [[ ! -f "$dependency_file" ]]; then
+        echo "ERROR: cache-key input does not exist: $dependency_file" >&2
+        exit 1
+    fi
+done
+
+CACHE_KEY=$(
+    {
+        printf 'rocm-arch=%s\n' "$ROCM_ARCH"
+        for dependency_file in "${dependency_files[@]}"; do
+            printf 'path=%s\n' "$dependency_file"
+            sha256sum "$dependency_file"
+        done
+    } | sha256sum | cut -c1-16
+)
+CACHE_REF="${REGISTRY}:rocm-deps-cache-${CACHE_KEY}"
+IMAGE_REF="${REGISTRY}:${BUILDKITE_COMMIT}"
+
+echo "ROCm dependency cache: ${CACHE_REF}"
+echo "ROCm CI image: ${IMAGE_REF}"
+
+# The docker-container driver supports exporting the full cache to a registry;
+# the default docker driver on CI does not.
+if ! docker buildx inspect "$BUILDER_NAME" >/dev/null 2>&1; then
+    docker buildx create --name "$BUILDER_NAME" --driver docker-container
+fi
+docker buildx use "$BUILDER_NAME"
+docker buildx inspect --bootstrap
+
+docker buildx build --push --pull --progress=plain \
+    --target test \
+    --build-arg "ARG_PYTORCH_ROCM_ARCH=${ROCM_ARCH}" \
+    --cache-from "type=registry,ref=${CACHE_REF}" \
+    --cache-to "type=registry,ref=${CACHE_REF},mode=max,compression=zstd,ignore-error=true" \
+    --file "$DOCKERFILE" \
+    --tag "$IMAGE_REF" .

.buildkite/amd/scripts/prewarm_cosmos3_aiter.py

@@ -0,0 +1,78 @@
+#!/usr/bin/env python3
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+
+"""Prebuild the BF16 AITER MHA modules used by Cosmos3 on ROCm."""
+
+from __future__ import annotations
+
+import os
+from pathlib import Path
+
+import torch
+
+from vllm_omni.diffusion.attention.backends.utils import fa
+from vllm_omni.platforms import current_omni_platform
+
+ATTENTION_VARIANTS = (
+    (True, "mha_fwd_bf16_nbias_mask_nlse_ndropout_nqscale"),
+    (False, "mha_fwd_bf16_nbias_nmask_nlse_ndropout_nqscale"),
+)
+
+
+def main() -> None:
+    if not current_omni_platform.is_rocm():
+        raise RuntimeError("The Cosmos3 AITER prewarm must run on ROCm")
+    if fa.flash_attn_func is None:
+        raise RuntimeError("ROCm FlashAttention/AITER is unavailable")
+
+    cache_dir_value = os.environ.get("AITER_JIT_DIR")
+    if not cache_dir_value:
+        raise RuntimeError("AITER_JIT_DIR must identify the job-local cache")
+    cache_dir = Path(cache_dir_value)
+
+    device = torch.device(current_omni_platform.device_type)
+    dtype = torch.bfloat16
+    batch_size = 1
+    sequence_length = 64
+    num_heads = 8
+    head_dim = 64
+
+    torch.manual_seed(42)
+    query = torch.randn(
+        batch_size,
+        sequence_length,
+        num_heads,
+        head_dim,
+        device=device,
+        dtype=dtype,
+    )
+    key = torch.randn_like(query)
+    value = torch.randn_like(query)
+
+    for causal, expected_module in ATTENTION_VARIANTS:
+        output = fa.flash_attn_func(
+            query,
+            key,
+            value,
+            dropout_p=0.0,
+            softmax_scale=head_dim**-0.5,
+            causal=causal,
+        )
+        if isinstance(output, tuple):
+            output = output[0]
+        current_omni_platform.synchronize()
+
+        if output.shape != query.shape:
+            raise RuntimeError(f"Unexpected AITER output shape: {output.shape}")
+        if not torch.isfinite(output).all():
+            raise RuntimeError("AITER prewarm produced non-finite output")
+
+        module_path = cache_dir / f"{expected_module}.so"
+        if not module_path.is_file():
+            raise RuntimeError(f"Expected AITER module was not built: {module_path}")
+        print(f"Prebuilt Cosmos3 AITER module: {module_path}", flush=True)
+
+
+if __name__ == "__main__":
+    main()

.buildkite/amd/scripts/select_test_suites.py

@@ -1,7 +1,7 @@
 #!/usr/bin/env python3
 # SPDX-License-Identifier: Apache-2.0
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
-"""Select AMD L2/L3 suites after Buildkite has created an AMD build.
+"""Select AMD ready/merge/nightly suites after Buildkite creates a build.
 
 This repository-side selector does not decide which GitHub label events start
 the external ``vllm-omni-amd-ci`` pipeline. Its Buildkite PR build condition
@@ -17,6 +17,7 @@ from collections.abc import Iterable
 SUITE_SPECS = {
     "ready": "READY_TESTS:test-amd-ready.yml",
     "merge": "MERGE_TESTS:test-amd-merge.yml",
+    "nightly": "NIGHTLY_TESTS:test-amd-nightly.yml",
 }
 
 
@@ -28,7 +29,7 @@ def _parse_debug_suites(value: str) -> tuple[str, ...]:
     seen: set[str] = set()
     for suite in suites:
         if suite not in SUITE_SPECS:
-            raise ValueError(f"DEBUG_TEST_YAML entries must be 'merge' or 'ready', got {suite!r}")
+            raise ValueError(f"DEBUG_TEST_YAML entries must be 'merge', 'nightly', or 'ready', got {suite!r}")
         if suite in seen:
             raise ValueError(f"duplicate DEBUG_TEST_YAML suite {suite!r}")
         seen.add(suite)
@@ -40,18 +41,27 @@ def select_amd_test_suites(
     branch: str,
     labels: Iterable[str],
     debug_test_yaml: str = "",
+    nightly: bool = False,
 ) -> tuple[str, ...]:
     """Return ordered suite names while preserving legacy unlabeled PRs."""
 
     if debug_test_yaml:
         return _parse_debug_suites(debug_test_yaml)
     if branch == "main":
-        return ("merge",)
+        return ("nightly",) if nightly else ("merge",)
 
     label_set = {label.strip() for label in labels if label.strip()}
-    selected = tuple(suite for label, suite in (("ready", "ready"), ("merge-test", "merge")) if label in label_set)
-    # AMD historically uploaded L2 for every PR build. Keep that behavior for
-    # triggers such as amd-test while adding tier selection for ready/merge-test.
+    selected = tuple(
+        suite
+        for label, suite in (
+            ("ready", "ready"),
+            ("merge-test", "merge"),
+            ("nightly-test", "nightly"),
+        )
+        if label in label_set
+    )
+    # AMD historically uploaded the ready suite for every admitted PR build.
+    # Preserve that fallback when no explicit tier label is present.
     return selected or ("ready",)
 
 
@@ -60,6 +70,7 @@ def main() -> int:
     parser.add_argument("--branch", required=True)
     parser.add_argument("--labels", default="")
     parser.add_argument("--debug-test-yaml", default="")
+    parser.add_argument("--nightly", type=int, choices=(0, 1), default=0)
     args = parser.parse_args()
 
     labels = tuple(args.labels.splitlines())
@@ -68,19 +79,15 @@ def main() -> int:
             branch=args.branch,
             labels=labels,
             debug_test_yaml=args.debug_test_yaml,
+            nightly=bool(args.nightly),
         )
     except ValueError as exc:
         parser.error(str(exc))
 
     label_set = {label.strip() for label in labels if label.strip()}
-    if not args.debug_test_yaml and args.branch != "main" and not ({"ready", "merge-test"} & label_set):
+    if args.branch != "main" and not args.debug_test_yaml and not ({"ready", "merge-test", "nightly-test"} & label_set):
         print(
-            "No AMD L2/L3 tier label found; preserving the legacy ready-suite fallback.",
-            file=sys.stderr,
-        )
-    if "nightly-test" in label_set:
-        print(
-            "AMD nightly-test has no suite yet and does not affect L2/L3 selection.",
+            "No AMD ready/merge/nightly label found; preserving the legacy ready-suite fallback.",
             file=sys.stderr,
         )
 

.buildkite/amd/test-amd-merge.yml

@@ -9,15 +9,26 @@ steps:
   steps:
     - label: "Simple · Model Executor Test · Shard %N/%t"
       agent_pool: mi300_1
-      parallelism: 2
+      parallelism: 3
       depends_on: amd-build
       mirror_hardwares: [amdproduction]
       grade: Blocking
+      timeout_in_minutes: 90
       commands:
         - export VLLM_ROCM_USE_AITER=0
         # ignore test_teacache_extractors.py because it use rocm gemm kernel from vLLM
         # that is not supported on CPU
-        - "pytest -sv tests/model_executor -m 'core_model and cpu' --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB"
+        - "pytest -sv tests/model_executor -m 'core_model and cpu and not omni' --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB"
+
+    - label: "Simple · Omni Processor Test"
+      agent_pool: mi300_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      timeout_in_minutes: 60
+      commands:
+        - export VLLM_ROCM_USE_AITER=0
+        - "pytest -sv tests/model_executor/models/test_omni_processing.py -m 'core_model and cpu and omni'"
 
     - label: "Simple · Diffusion Test · Shard %N/%t"
       agent_pool: mi300_1
@@ -25,11 +36,36 @@ steps:
       depends_on: amd-build
       mirror_hardwares: [amdproduction]
       grade: Blocking
+      timeout_in_minutes: 45
+      # Buildkite's status-based rule cannot prove every exit 134 is a GPU hang.
+      # It retries 134 once; the Kubernetes job retry starts a fresh pod.
+      # Assertions (1), timeouts (124), and OOMs (137) remain non-retryable.
+      retry:
+        automatic:
+          - exit_status: 134
+            limit: 1
       commands:
         - export VLLM_ROCM_USE_AITER=0
         # ignore test_teacache_extractors.py because it use rocm gemm kernel from vLLM
         # that is not supported on CPU
-        - "pytest -sv tests/diffusion -m 'core_model and cpu' --ignore=tests/diffusion/cache/test_teacache_extractors.py --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB"
+        # Keep multi-GPU tests out of this one-GPU pool even when their module
+        # also carries the cpu marker for unit tests.
+        # Keep π0.5 CPU units out of these shards: CUDA READY/MERGE already run
+        # the same suite mandatorily, and AMD covers it non-blocking in nightly.
+        - "pytest -sv tests/diffusion -m 'core_model and cpu and not (cards_2 or cards_3 or cards_4 or cards_5 or cards_6 or cards_7 or cards_8)' --ignore=tests/diffusion/cache/test_teacache_extractors.py --ignore=tests/diffusion/models/pi05/test_pi05_units.py --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB"
+
+    - label: "Diffusion · AuK VAE GPU Test"
+      agent_pool: mi300_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      timeout_in_minutes: 40
+      commands:
+        - export VLLM_ROCM_USE_AITER=0
+        - >-
+          timeout --signal=TERM --kill-after=2m 35m
+          pytest -sv tests/diffusion/models/auk/test_auk_vae_fastpath.py
+          -m 'core_model and rocm and cards_1' --run-level core_model
 
     - label: "Simple · Engine&Entrypoints Test"
       agent_pool: mi300_1
@@ -53,6 +89,25 @@ steps:
 
 - group: ":card_index_dividers: Diffusion Test"
   steps:
+    # AR paged attention is real GPU work, including cold AITER compilation.
+    # Keep its complete numerical matrix out of the CPU-marked Simple shards.
+    - label: "ROCm · AR Diffusion Paged Attention GPU Test"
+      agent_pool: mi300_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      timeout_in_minutes: 20
+      commands:
+        - export VLLM_ROCM_USE_AITER=1
+        - export VLLM_OMNI_AR_FA_REQUIRED=1
+        - export AITER_JIT_DIR="/tmp/vllm-omni-ar-aiter-$$BUILDKITE_JOB_ID"
+        - export TORCH_EXTENSIONS_DIR="/tmp/vllm-omni-ar-torch-extensions-$$BUILDKITE_JOB_ID"
+        - mkdir -p "$$AITER_JIT_DIR" "$$TORCH_EXTENSIONS_DIR"
+        - >-
+          timeout --signal=TERM --kill-after=2m 15m
+          pytest -sv tests/diffusion/ar_diffusion/test_paged_attention.py
+          -m 'core_model and rocm and MI325 and cards_1' --run-level core_model
+
     - label: "Diffusion · LoRA Test"
       agent_pool: mi300_1
       depends_on: amd-build
@@ -71,6 +126,7 @@ steps:
       commands:
         - export VLLM_LOGGING_LEVEL=DEBUG
         - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+        - export DIFFUSION_ATTENTION_BACKEND=TORCH_SDPA
         - pytest -s -v tests/diffusion/offloader/test_diffusion_cpu_offload.py
         - >-
           pytest -s -v tests/diffusion/offloader/test_diffusion_layerwise_offload.py
@@ -119,6 +175,11 @@ steps:
       grade: Blocking
       commands:
         - pytest -s -v tests/diffusion/distributed/test_tensor_parallel.py
+        - >-
+          timeout 20m pytest -s -v
+          tests/diffusion/attention/test_ulysses_uaa.py::test_ulysses_uaa_2d_mask_layout_matches_baseline
+          -m 'core_model and cards_2'
+          --run-level "core_model"
 
     - label: "Diffusion Model Test"
       agent_pool: mi300_1
@@ -126,7 +187,9 @@ steps:
       mirror_hardwares: [amdproduction]
       grade: Blocking
       commands:
-        - timeout 30m pytest -s -v tests/e2e/offline_inference/test_z_image.py
+        # A cold Z-Image run spends about six minutes loading weights before
+        # the first-request AITER compilation, which can itself exceed 23m.
+        - timeout 55m pytest -s -v tests/e2e/offline_inference/test_z_image.py
 
     - label: "Diffusion GPU Worker Test"
       agent_pool: mi300_2

.buildkite/amd/test-amd-nightly.yml

@@ -0,0 +1,170 @@
+# AMD MI300 `full_model` nightly coverage during burn-in.
+# All jobs are non-blocking until the lane has stable runtime evidence.
+# Historical build #11646 was red in Function Expansion and Accuracy (one
+# async_chunk similarity mismatch; Seed-TTS WER 0.376480 > 0.35).
+
+env:
+  VLLM_WORKER_MULTIPROC_METHOD: spawn
+  HF_HUB_DOWNLOAD_TIMEOUT: 300
+  HF_HUB_ETAG_TIMEOUT: 60
+
+steps:
+  # MI325 is the repository's legacy ROCm pytest marker. The AMD template maps
+  # agent_pool to the current MI300 Buildkite queues independently of that mark.
+  - group: ":full_moon: Qwen3-Omni MI300 Nightly"
+    steps:
+      - label: "Qwen3-Omni Function Expansion"
+        agent_pool: mi300_2
+        depends_on: amd-build
+        mirror_hardwares: [amdproduction]
+        grade: NonBlocking
+        timeout_in_minutes: 180
+        commands:
+          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+          - pytest -s -v tests/e2e/online_serving/test_qwen3_omni_expansion.py -m "full_model and rocm and MI325 and cards_2" --run-level "full_model"
+
+      - label: "Qwen3-Omni Accuracy"
+        agent_pool: mi300_2
+        depends_on: amd-build
+        mirror_hardwares: [amdproduction]
+        grade: NonBlocking
+        timeout_in_minutes: 180
+        artifact_paths:
+          - tests/e2e/accuracy/qwen3_omni/results/qwen_omni_acc/*.json
+        commands:
+          - export SEED_TTS_WER_EVAL=1
+          # PyTorch ROCm uses the torch.cuda device namespace.
+          - export SEED_TTS_EVAL_DEVICE=cuda:1
+          - pytest -s -v tests/e2e/accuracy/qwen3_omni/test_qwen3_omni.py -m "full_model and rocm and MI325 and cards_2" --run-level "full_model"
+          # Native tests run from the image copy under /app/vllm-omni, while
+          # Buildkite's artifact hook searches the checkout workspace.
+          - |
+            if [[ "$$PWD" != "$${BUILDKITE_BUILD_CHECKOUT_PATH:?}" ]]; then
+              artifact_dir="$${BUILDKITE_BUILD_CHECKOUT_PATH}/tests/e2e/accuracy/qwen3_omni/results/qwen_omni_acc"
+              mkdir -p "$$artifact_dir"
+              cp -v tests/e2e/accuracy/qwen3_omni/results/qwen_omni_acc/*.json "$$artifact_dir/"
+            fi
+
+      - label: "Qwen3-Omni Documentation Examples"
+        agent_pool: mi300_2
+        depends_on: amd-build
+        mirror_hardwares: [amdproduction]
+        grade: NonBlocking
+        timeout_in_minutes: 120
+        commands:
+          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+          - pytest -s -v tests/examples/online_serving/test_qwen3_omni.py -m "full_model and rocm and MI325 and cards_2" --run-level "full_model"
+
+      - label: "Qwen3-Omni AITER-on Smoke"
+        agent_pool: mi300_2
+        depends_on: amd-build
+        mirror_hardwares: [amdproduction]
+        grade: NonBlocking
+        timeout_in_minutes: 90
+        commands:
+          - export VLLM_ROCM_USE_AITER=1
+          - pytest -s -v tests/e2e/online_serving/test_qwen3_omni_expansion.py::test_text_to_text_audio_001 -m "full_model and rocm and MI325 and cards_2" --run-level "full_model"
+
+  - group: ":speaker: CosyVoice3 MI300 Nightly"
+    steps:
+      # Exercise all sync/async and Chinese/English cases with the platform's
+      # default attention backend. On ROCm this retains end-to-end AITER
+      # coverage without putting its cold JIT build on the READY critical path.
+      - label: "CosyVoice3-TTS E2E Test"
+        agent_pool: mi300_1
+        depends_on: amd-build
+        mirror_hardwares: [amdproduction]
+        grade: NonBlocking
+        timeout_in_minutes: 90
+        retry:
+          automatic:
+            - exit_status: 134
+              limit: 1
+        commands:
+          - pytest -s -v tests/e2e/online_serving/test_cosyvoice3_tts_expansion.py -m "slow" --run-level "core_model"
+
+  # π0.5 CPU units are mandatory on CUDA READY/MERGE. On AMD they exercise the
+  # ROCm container/import path with CPU tensors only, so keep them off the
+  # blocking READY/MERGE critical path and retain non-blocking nightly coverage.
+  - group: ":robot_face: Pi0.5 CPU Nightly"
+    steps:
+      - label: "Simple · Pi0.5 CPU Test"
+        agent_pool: mi300_1
+        mirror_hardwares: [amdproduction]
+        grade: NonBlocking
+        timeout_in_minutes: 60
+        commands:
+          - export VLLM_ROCM_USE_AITER=0
+          - "pytest -sv tests/diffusion/models/pi05/test_pi05_units.py -m 'core_model and cpu'"
+
+  - group: ":full_moon: LingBot-Video MI300 Nightly"
+    steps:
+      - label: "LingBot-Video Dense CFG-off Smoke"
+        agent_pool: mi300_1
+        mirror_hardwares: [amdproduction]
+        grade: NonBlocking
+        timeout_in_minutes: 90
+        artifact_paths:
+          - artifacts/rocm-lingbot-nightly/**/*
+        commands:
+          - pytest --collect-only -q tests/e2e/online_serving/test_lingbot_video_expansion.py::test_cfg_off -m "full_model and diffusion and rocm and MI325 and cards_1" --run-level full_model
+          - pytest -s -v tests/e2e/online_serving/test_lingbot_video_expansion.py::test_cfg_off -m "full_model and diffusion and rocm and MI325 and cards_1" --run-level full_model --junitxml=/tmp/lingbot-pytest.xml
+          - |
+            artifact_dir="$${BUILDKITE_BUILD_CHECKOUT_PATH:?}/artifacts/rocm-lingbot-nightly"
+            mkdir -p "$$artifact_dir"
+            cp -v /tmp/lingbot-pytest.xml "$$artifact_dir/"
+
+  - group: ":full_moon: Cosmos3 MI300 Nightly"
+    steps:
+      - label: "Cosmos3 T2I Function"
+        agent_pool: mi300_1
+        mirror_hardwares: [amdproduction]
+        grade: NonBlocking
+        timeout_in_minutes: 90
+        artifact_paths:
+          - artifacts/rocm-cosmos3-nightly/**/*
+        commands:
+          # Cosmos3 uses both masked and unmasked BF16 AITER MHA modules. Build
+          # them before server startup, then reuse the same job-local cache
+          # during the smoke test.
+          - export VLLM_ROCM_USE_AITER=1
+          - export AITER_JIT_DIR="/tmp/vllm-omni-cosmos3-aiter-$$BUILDKITE_JOB_ID"
+          - export TORCH_EXTENSIONS_DIR="/tmp/vllm-omni-cosmos3-torch-extensions-$$BUILDKITE_JOB_ID"
+          - mkdir -p "$$AITER_JIT_DIR" "$$TORCH_EXTENSIONS_DIR"
+          - pytest --collect-only -q tests/e2e/online_serving/test_cosmos3.py::test_text_to_image_001 -m "core_model and rocm and MI325 and cards_1" --run-level full_model
+          - >-
+            timeout --signal=TERM --kill-after=2m 55m
+            python .buildkite/amd/scripts/prewarm_cosmos3_aiter.py
+          - >-
+            timeout --signal=TERM --kill-after=2m 25m
+            pytest -s -v
+            tests/e2e/online_serving/test_cosmos3.py::test_text_to_image_001
+            -m "core_model and rocm and MI325 and cards_1"
+            --run-level full_model
+            --junitxml=/tmp/cosmos3-pytest.xml
+          - |
+            artifact_dir="$${BUILDKITE_BUILD_CHECKOUT_PATH:?}/artifacts/rocm-cosmos3-nightly"
+            mkdir -p "$$artifact_dir"
+            cp -v /tmp/cosmos3-pytest.xml "$$artifact_dir/"
+
+  - group: ":full_moon: HunyuanImage3 MI300 Nightly"
+    steps:
+      - label: "HunyuanImage3 Offline Pixel Accuracy"
+        agent_pool: mi300_4
+        mirror_hardwares: [amdproduction]
+        grade: NonBlocking
+        timeout_in_minutes: 180
+        env:
+          HUNYUAN_IMAGE3_MODEL: "tencent/HunyuanImage-3.0-Instruct"
+          HUNYUAN_IMAGE3_DEVICES: "0,1,2,3"
+          DIFFUSION_ATTENTION_BACKEND: TORCH_SDPA
+        artifact_paths:
+          - artifacts/rocm-hunyuanimage3/**/*
+        commands:
+          - pytest --collect-only -q tests/e2e/accuracy/test_hunyuan_image3_pixel_accuracy.py::test_hunyuan_image3_pixel_accuracy_offline -m "full_model and rocm and MI325 and cards_4" --run-level full_model
+          - pytest -s -v tests/e2e/accuracy/test_hunyuan_image3_pixel_accuracy.py::test_hunyuan_image3_pixel_accuracy_offline -m "full_model and rocm and MI325 and cards_4" --run-level full_model --junitxml=/tmp/hunyuanimage3-pytest.xml
+          - |
+            artifact_dir="$${BUILDKITE_BUILD_CHECKOUT_PATH:?}/artifacts/rocm-hunyuanimage3"
+            mkdir -p "$$artifact_dir"
+            cp -v tests/e2e/accuracy/artifacts/HunyuanImage-3_0-Instruct/*.png "$$artifact_dir/"
+            cp -v /tmp/hunyuanimage3-pytest.xml "$$artifact_dir/"

.buildkite/amd/test-amd-ready.yml

@@ -9,26 +9,63 @@ steps:
   steps:
     - label: "Simple · Model Executor Test · Shard %N/%t"
       agent_pool: mi300_1
-      parallelism: 2
+      parallelism: 3
       depends_on: amd-build
       mirror_hardwares: [amdproduction]
       grade: Blocking
+      timeout_in_minutes: 90
       commands:
         - export VLLM_ROCM_USE_AITER=0
         # ignore test_teacache_extractors.py because it use rocm gemm kernel from vLLM
         # that is not supported on CPU
-        - "pytest -sv tests/model_executor -m 'core_model and cpu' --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB"
+        - "pytest -sv tests/model_executor -m 'core_model and cpu and not omni' --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB"
 
-    - label: "Simple · Diffusion Test"
+    - label: "Simple · Omni Processor Test"
       agent_pool: mi300_1
       depends_on: amd-build
       mirror_hardwares: [amdproduction]
       grade: Blocking
+      timeout_in_minutes: 60
+      commands:
+        - export VLLM_ROCM_USE_AITER=0
+        - "pytest -sv tests/model_executor/models/test_omni_processing.py -m 'core_model and cpu and omni'"
+
+    - label: "Simple · Diffusion Test · Shard %N/%t"
+      agent_pool: mi300_1
+      parallelism: 4
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      timeout_in_minutes: 45
+      # Buildkite's status-based rule cannot prove every exit 134 is a GPU hang.
+      # It retries 134 once; the Kubernetes job retry starts a fresh pod.
+      # Assertions (1), timeouts (124), and OOMs (137) remain non-retryable.
+      retry:
+        automatic:
+          - exit_status: 134
+            limit: 1
       commands:
         - export VLLM_ROCM_USE_AITER=0
         # ignore test_teacache_extractors.py because it use rocm gemm kernel from vLLM
         # that is not supported on CPU
-        - "pytest -sv tests/diffusion -m 'core_model and cpu'  --ignore=tests/diffusion/cache/test_teacache_extractors.py"
+        # Keep multi-GPU tests out of this one-GPU pool even when their module
+        # also carries the cpu marker for unit tests.
+        # Keep π0.5 CPU units out of these shards: CUDA READY/MERGE already run
+        # the same suite mandatorily, and AMD covers it non-blocking in nightly.
+        - "pytest -sv tests/diffusion -m 'core_model and cpu and not (cards_2 or cards_3 or cards_4 or cards_5 or cards_6 or cards_7 or cards_8)' --ignore=tests/diffusion/cache/test_teacache_extractors.py --ignore=tests/diffusion/models/pi05/test_pi05_units.py --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB"
+
+    - label: "Diffusion · AuK VAE GPU Test"
+      agent_pool: mi300_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      timeout_in_minutes: 40
+      commands:
+        - export VLLM_ROCM_USE_AITER=0
+        - >-
+          timeout --signal=TERM --kill-after=2m 35m
+          pytest -sv tests/diffusion/models/auk/test_auk_vae_fastpath.py
+          -m 'core_model and rocm and cards_1' --run-level core_model
 
     - label: "Simple · Engine&Entrypoints Test"
       agent_pool: mi300_1
@@ -69,6 +106,55 @@ steps:
 
 - group: ":card_index_dividers: Diffusion Test"
   steps:
+    # Keep real FlashAttention execution out of the CPU-marked Simple lane.
+    # AITER may need to build several kernels on a cold worker, so run one
+    # pytest worker with job-local build directories and a bounded timeout.
+    - label: "ROCm · FlashAttention/AITER GPU Coverage"
+      agent_pool: mi300_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: NonBlocking
+      timeout_in_minutes: 60
+      commands:
+        - export VLLM_ROCM_USE_AITER=1
+        - export AITER_JIT_DIR="/tmp/vllm-omni-aiter-$$BUILDKITE_JOB_ID"
+        - export TORCH_EXTENSIONS_DIR="/tmp/vllm-omni-torch-extensions-$$BUILDKITE_JOB_ID"
+        - mkdir -p "$$AITER_JIT_DIR" "$$TORCH_EXTENSIONS_DIR"
+        - >-
+          timeout --signal=TERM --kill-after=2m 55m
+          pytest -s -v -n 1
+          tests/diffusion/attention/
+          -m 'core_model and rocm and MI325 and cards_1'
+          --run-level "core_model"
+        # MammothModa2 exercises the same two AITER MHA modules built by the
+        # preceding suite. Keep its default-backend run in this job so a cold
+        # worker compiles each module once instead of timing out the generic
+        # model job.
+        - >-
+          timeout --signal=TERM --kill-after=2m 10m
+          pytest -s -v -n 1
+          tests/diffusion/models/mammoth_moda2/test_dit_attention_cuda.py
+          --run-level "core_model"
+
+    # AR paged attention is real GPU work, including cold AITER compilation.
+    # Keep its complete numerical matrix out of the CPU-marked Simple shards.
+    - label: "ROCm · AR Diffusion Paged Attention GPU Test"
+      agent_pool: mi300_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      timeout_in_minutes: 20
+      commands:
+        - export VLLM_ROCM_USE_AITER=1
+        - export VLLM_OMNI_AR_FA_REQUIRED=1
+        - export AITER_JIT_DIR="/tmp/vllm-omni-ar-aiter-$$BUILDKITE_JOB_ID"
+        - export TORCH_EXTENSIONS_DIR="/tmp/vllm-omni-ar-torch-extensions-$$BUILDKITE_JOB_ID"
+        - mkdir -p "$$AITER_JIT_DIR" "$$TORCH_EXTENSIONS_DIR"
+        - >-
+          timeout --signal=TERM --kill-after=2m 15m
+          pytest -sv tests/diffusion/ar_diffusion/test_paged_attention.py
+          -m 'core_model and rocm and MI325 and cards_1' --run-level core_model
+
     - label: "Diffusion · Batch Test"
       agent_pool: mi300_1
       depends_on: amd-build
@@ -107,6 +193,11 @@ steps:
         - export VLLM_LOGGING_LEVEL=DEBUG
         - export VLLM_WORKER_MULTIPROC_METHOD=spawn
         - timeout 20m pytest -s -v tests/diffusion/distributed/test_sequence_parallel.py -m core_model
+        - >-
+          timeout 20m pytest -s -v
+          tests/diffusion/attention/test_ulysses_uaa.py::test_ulysses_uaa_2d_mask_layout_matches_baseline
+          -m 'core_model and cards_2'
+          --run-level "core_model"
         - >-
           timeout 20m pytest -s -v
           tests/diffusion/models/ltx2/test_ltx2_transformer_ulysses.py
@@ -131,9 +222,18 @@ steps:
         # registered multi-GPU marker from this one-GPU job.
         - >-
           timeout 15m pytest -sv tests/diffusion/models/
+          --ignore=tests/diffusion/models/auk/test_auk_vae_fastpath.py
+          --ignore=tests/diffusion/models/mammoth_moda2/test_dit_attention_cuda.py
           -m 'core_model and cuda and not
           (cards_2 or cards_3 or cards_4 or cards_5 or cards_6 or cards_7 or cards_8)'
           --run-level "core_model"
+        # Keep the model-specific numerical coverage blocking without paying
+        # the default ROCm backend's cold AITER compilation cost in this lane.
+        - >-
+          timeout 10m env DIFFUSION_ATTENTION_BACKEND=TORCH_SDPA
+          pytest -sv
+          tests/diffusion/models/mammoth_moda2/test_dit_attention_cuda.py
+          --run-level "core_model"
 
     - label: "Diffusion · Wan22 Test"
       agent_pool: mi300_1
@@ -160,6 +260,38 @@ steps:
 #     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
 #     - timeout 20m pytest -s -v tests/e2e/offline_inference/test_stable_audio_expansion.py -m "advanced_model and diffusion and L4" --run-level advanced_model
 
+- label: "ROCm · Model Executor GPU Coverage (R2-01)"
+  agent_pool: mi300_1
+  depends_on: amd-build
+  mirror_hardwares: [amdproduction]
+  grade: NonBlocking
+  timeout_in_minutes: 20
+  artifact_paths:
+    - "artifacts/rocm-r2-01/**/*"
+  commands:
+    # Native AMD tests run from /app/vllm-omni, while Buildkite uploads
+    # artifacts relative to its checkout directory.
+    - export R2_01_ARTIFACT_DIR="$$BUILDKITE_BUILD_CHECKOUT_PATH/artifacts/rocm-r2-01"
+    - mkdir -p "$$R2_01_ARTIFACT_DIR"
+    # The dedicated Engine Test below owns the only selected tests/engine case;
+    # do not duplicate its model startup in this time-bounded coverage lane.
+    # The CUDA ready job also covers tests/worker/test_batched_omni_output.py.
+    # Keep the initial ROCm burn-in scoped to tests/model_executor/.
+    # Verbose pytest output and JUnit retain node IDs without a second
+    # collection-only pass over the same GPU-backed test tree.
+    - >-
+      pytest -s -v -ra
+      tests/model_executor/
+      -m 'core_model and cuda and not nvidia_only and not
+      (cards_2 or cards_3 or cards_4 or cards_5 or cards_6 or cards_7 or cards_8)'
+      --run-level "core_model"
+      --durations=0
+      --junitxml="$$R2_01_ARTIFACT_DIR/pytest.xml"
+      2>&1 | tee "$$R2_01_ARTIFACT_DIR/pytest.log"
+    - >-
+      tail -n 100 "$$R2_01_ARTIFACT_DIR/pytest.log" >
+      "$$R2_01_ARTIFACT_DIR/pytest-summary.txt"
+
 - label: "Engine Test"
   agent_pool: mi300_1
   depends_on: amd-build
@@ -172,6 +304,71 @@ steps:
               pytest -s -v tests/engine/test_async_omni_engine_abort.py
       '
 
+- label: "ROCm · Entrypoints GPU Coverage (R2-02)"
+  agent_pool: mi300_1
+  depends_on: amd-build
+  mirror_hardwares: [amdproduction]
+  grade: NonBlocking
+  timeout_in_minutes: 30
+  env:
+    VLLM_WORKER_MULTIPROC_METHOD: spawn
+  artifact_paths:
+    - "artifacts/rocm-r2-02/**/*"
+  commands:
+    # Native AMD tests run from /app/vllm-omni, while Buildkite uploads
+    # artifacts relative to its checkout directory.
+    - export R2_02_ARTIFACT_DIR="$$BUILDKITE_BUILD_CHECKOUT_PATH/artifacts/rocm-r2-02"
+    - mkdir -p "$$R2_02_ARTIFACT_DIR"
+    - >-
+      ps -eo pid=,ppid=,lstart=,stat=,comm= >
+      "$$R2_02_ARTIFACT_DIR/processes-before.txt"
+    # Verbose pytest output and JUnit retain selected node IDs without a
+    # second collection-only pass over the same GPU-backed test tree.
+    # The inner timeout leaves five minutes for teardown verification and
+    # artifact generation before the Buildkite step timeout.
+    - >-
+      timeout --signal=TERM --kill-after=1m 25m
+      pytest -s -v -ra
+      tests/entrypoints/
+      -m 'core_model and cuda and not
+      (cards_2 or cards_3 or cards_4 or cards_5 or cards_6 or cards_7 or cards_8)'
+      --run-level "core_model"
+      --durations=0
+      --junitxml="$$R2_02_ARTIFACT_DIR/pytest.xml"
+      2>&1 | tee "$$R2_02_ARTIFACT_DIR/pytest.log"
+    - >-
+      tail -n 100 "$$R2_02_ARTIFACT_DIR/pytest.log" >
+      "$$R2_02_ARTIFACT_DIR/pytest-summary.txt"
+    - |
+      sleep 5
+      ps -eo pid=,ppid=,lstart=,stat=,comm= > "$$R2_02_ARTIFACT_DIR/processes-after.txt"
+      # Compare PID plus process start time so PID reuse cannot hide a leak.
+      # Double dollars preserve awk field references through Buildkite interpolation.
+      awk '
+        NR == FNR {
+          if ($$9 != "ps") {
+            baseline[$$1 "|" $$3 "|" $$4 "|" $$5 "|" $$6 "|" $$7] = 1
+          }
+          next
+        }
+        $$9 != "ps" {
+          identity = $$1 "|" $$3 "|" $$4 "|" $$5 "|" $$6 "|" $$7
+          if (!(identity in baseline)) {
+            print
+            leaked = 1
+          }
+        }
+        END {
+          if (leaked) {
+            print "status=FAIL leaked_processes_detected=1"
+            exit 1
+          }
+          print "status=PASS leaked_processes=0"
+        }
+      ' "$$R2_02_ARTIFACT_DIR/processes-before.txt" \
+        "$$R2_02_ARTIFACT_DIR/processes-after.txt" \
+        2>&1 | tee "$$R2_02_ARTIFACT_DIR/process-cleanup.txt"
+
 # - label: "Omni Model Test Qwen3-Omni"
 #   agent_pool: mi300_2
 #   depends_on: amd-build
@@ -196,13 +393,23 @@ steps:
         - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
         - timeout 30m pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py -m "core_model" --run-level "core_model"
 
-    - label: "CosyVoice3-TTS E2E Test"
+    # Keep a fast model/server integration check in the blocking lane. The
+    # full default-backend suite remains in the nightly pipeline because a
+    # cold ROCm AITER MHA build can add more than 25 minutes to this job.
+    - label: "CosyVoice3-TTS E2E Smoke (SDPA)"
       agent_pool: mi300_1
       depends_on: amd-build
       mirror_hardwares: [amdproduction]
       grade: Blocking
+      # A ROCm hardware hang aborts the interpreter with 134. Let Buildkite
+      # create one fresh pod instead of retrying on the poisoned GPU process.
+      retry:
+        automatic:
+          - exit_status: 134
+            limit: 1
       commands:
-        - pytest -s -v tests/e2e/online_serving/test_cosyvoice3_tts_expansion.py -m "slow" --run-level "core_model"
+        - export DIFFUSION_ATTENTION_BACKEND=TORCH_SDPA
+        - pytest -s -v tests/e2e/online_serving/test_cosyvoice3_tts_expansion.py::test_voice_clone_zh_002 -m "slow" --run-level "core_model"
 
 # TODO: Bagel test on ROCm is very unstable. @tjtanaa
 # Need to debug before reneable numerical changes across large PRs

.buildkite/amd/test-template-amd-omni.j2

@@ -3,7 +3,8 @@
    Last synced: 2026-08-04
    Modifications: Use the shared MI300 queues and Omni command-passing contract
 #}
-{% set docker_image_amd = "rocm/vllm-omni:$BUILDKITE_COMMIT" %}
+{% set docker_registry_amd = "rocm/vllm-omni" %}
+{% set docker_image_amd = docker_registry_amd ~ ":$BUILDKITE_COMMIT" %}
 {% set default_working_dir = "/app/vllm-omni" %}
 
 {# MI300 runs tests natively in Kubernetes; Docker-in-Docker is unavailable. #}
@@ -74,11 +75,12 @@ steps:
         depends_on: ~
         soft_fail: false
         commands:
-          - "docker build -f docker/Dockerfile.rocm -t {{ docker_image_amd }} --target test --build-arg ARG_PYTORCH_ROCM_ARCH=gfx942 --progress plain ."
-          - "docker push {{ docker_image_amd }}"
+          - "bash .buildkite/amd/scripts/build-ci-image.sh"
         key: "amd-build"
         env:
           DOCKER_BUILDKIT: "1"
+          PYTORCH_ROCM_ARCH: "gfx942"
+          ROCM_CI_IMAGE_REGISTRY: "{{ docker_registry_amd }}"
         retry:
           automatic:
             - exit_status: -1  # Agent was lost
@@ -156,6 +158,14 @@ steps:
 {% for path in step.artifact_paths %}
           - "{{ path }}"
 {% endfor %}
+{% endif %}
+{% if step.retry %}
+        retry:
+          automatic:
+{% for retry_rule in step.retry.automatic %}
+            - exit_status: {{ retry_rule.exit_status }}
+              limit: {{ retry_rule.limit }}
+{% endfor %}
 {% endif %}
         {% if step.timeout_in_minutes %}
         timeout_in_minutes: {{ step.timeout_in_minutes }}
@@ -224,6 +234,14 @@ steps:
 {% for path in step.artifact_paths %}
       - "{{ path }}"
 {% endfor %}
+{% endif %}
+{% if step.retry %}
+    retry:
+      automatic:
+{% for retry_rule in step.retry.automatic %}
+        - exit_status: {{ retry_rule.exit_status }}
+          limit: {{ retry_rule.limit }}
+{% endfor %}
 {% endif %}
     {% if step.timeout_in_minutes %}
     timeout_in_minutes: {{ step.timeout_in_minutes }}

.buildkite/common/ci_source_file_dependencies.yml

@@ -0,0 +1,525 @@
+# Uploader-only path-prefix presets for CUDA L2–L5 and NPU L4 test YAML.
+# Referenced as ``source_file_dependencies: omni_qwen3_omni_function`` (or a list of
+# keys) on a step; ``upload_pipeline.py`` expands the key(s) only.
+#
+# YAML anchors above the registry hold model business-code paths. Job keys
+# ({model}_function / _perf / _accuracy / _reliability / _doc / _cov) alias
+# those anchors and add the scripts the job actually runs. Feature keys below
+# are unchanged. ``source_filter_fallback`` is not a job key: when no job-key
+# prefix matches, a matching diff keeps every job for that upload.
+#
+# YAML anchors are nested lists; the loader flattens them.
+
+x-qwen3-tts: &qwen3_tts
+  - vllm_omni/engine/async_omni_engine.py
+  - vllm_omni/entrypoints/omni_base.py
+  - vllm_omni/worker/gpu_model_runner.py
+  - vllm_omni/model_executor/models/qwen3_tts/
+  - vllm_omni/model_executor/models/common/qwen3_code_predictor.py
+  - vllm_omni/model_executor/models/common/fused_sampling.py
+  - vllm_omni/model_executor/models/common/short_kv_attention.py
+  - vllm_omni/model_executor/models/common/talker_first_audio.py
+  - vllm_omni/model_executor/models/common/snake_activation.py
+  - vllm_omni/model_executor/models/common/alias_free_activation.py
+  - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
+  - vllm_omni/deploy/qwen3_tts.yaml
+  - vllm_omni/deploy/qwen3_tts_high_concurrency_mrv2_single_gpu.yaml
+  - vllm_omni/entrypoints/openai/serving_speech.py
+  - vllm_omni/entrypoints/openai/tts_adapters/
+  - vllm_omni/engine/stage_pool.py
+  - vllm_omni/worker_v2/
+  - vllm_omni/core/sched/
+  - vllm_omni/distributed/omni_connectors/
+  - vllm_omni/config/stage_config.py
+  - vllm_omni/config/omni_config.py
+  - vllm_omni/engine/orchestrator.py
+  - vllm_omni/engine/stage_engine_core_proc.py
+  - vllm_omni/engine/cuda_mps.py
+  - vllm_omni/engine/stage_runtime.py
+  - vllm_omni/engine/stage_init_utils.py
+  - vllm_omni/deploy/qwen3_tts_mrv2.yaml
+  - vllm_omni/deploy/qwen3_tts_high_concurrency_mrv2.yaml
+  - vllm_omni/deploy/qwen3_tts_high_concurrency_mrv2_b4.yaml
+
+x-minicpmo-4-5: &minicpmo_4_5
+  - vllm_omni/engine/async_omni_engine.py
+  - vllm_omni/entrypoints/omni_base.py
+  - vllm_omni/model_executor/models/minicpmo_4_5/
+  - vllm_omni/model_executor/models/step_audio2/
+  - vllm_omni/model_executor/models/cosyvoice3/code2wav_core/
+  - vllm_omni/model_executor/models/cosyvoice3/utils.py
+  - vllm_omni/model_executor/stage_input_processors/minicpmo_4_5_omni.py
+  - vllm_omni/deploy/minicpmo_4_5.yaml
+  - vllm_omni/transformers_utils/chat_templates/minicpmo45_native.jinja
+
+x-minicpmo-4-5-duplex: &minicpmo_4_5_duplex
+  - vllm_omni/deploy/minicpmo_4_5_duplex.yaml
+  - vllm_omni/engine/duplex/
+  - vllm_omni/engine/duplex_omni_engine.py
+  - vllm_omni/engine/duplex_orchestrator.py
+  - vllm_omni/engine/omni_engine_base.py
+  - vllm_omni/entrypoints/duplex/
+  - vllm_omni/entrypoints/duplex_omni.py
+  - vllm_omni/entrypoints/async_omni_base.py
+  - vllm_omni/protocol/
+  - vllm_omni/clients/
+  - vllm_omni/outputs/duplex.py
+  - vllm_omni/model_executor/duplex_sampling.py
+  - vllm_omni/experimental/fullduplex/
+  - vllm_omni/core/sched/
+  - vllm_omni/outputs/multimodal_accumulation.py
+  - vllm_omni/engine/stage_init_utils.py
+  - vllm_omni/engine/orchestrator.py
+  - vllm_omni/entrypoints/cli/serve.py
+  - vllm_omni/entrypoints/openai/api_server.py
+  - vllm_omni/entrypoints/openai/duplex_capability.py
+  - vllm_omni/config/stage_config.py
+
+x-qwen2-5-omni: &qwen2_5_omni
+  - vllm_omni/engine/async_omni_engine.py
+  - vllm_omni/entrypoints/omni_base.py
+  - vllm_omni/model_executor/models/qwen2_5_omni/
+  - vllm_omni/model_executor/stage_input_processors/qwen2_5_omni.py
+  - vllm_omni/deploy/qwen2_5_omni.yaml
+
+x-qwen3-omni: &qwen3_omni
+  - vllm_omni/engine/async_omni_engine.py
+  - vllm_omni/entrypoints/omni_base.py
+  - vllm_omni/entrypoints/openai/serving_chat.py
+  - vllm_omni/model_executor/models/qwen3_omni/
+  - vllm_omni/model_executor/models/common/qwen3_code_predictor.py
+  - vllm_omni/model_executor/models/common/fused_sampling.py
+  - vllm_omni/model_executor/models/common/short_kv_attention.py
+  - vllm_omni/model_executor/models/common/talker_first_audio.py
+  - vllm_omni/model_executor/models/common/snake_activation.py
+  - vllm_omni/model_executor/models/qwen2_5_omni/qwen2_5_omni_thinker.py
+  - vllm_omni/model_executor/models/qwen3_tts/cuda_graph_decoder_wrapper.py
+  - vllm_omni/model_executor/stage_input_processors/qwen3_omni.py
+  - vllm_omni/deploy/qwen3_omni_moe.yaml
+  - vllm_omni/deploy/qwen3_omni_moe_mrv2.yaml
+  - vllm_omni/deploy/qwen3_omni_moe_mrv2_h200.yaml
+  - vllm_omni/deploy/qwen3_omni_moe_thinking.yaml
+
+x-wan22: &wan22
+  - vllm_omni/diffusion/models/wan2_2/
+  - vllm_omni/diffusion/models/dmd2/
+
+x-cosmos3: &cosmos3
+  - vllm_omni/diffusion/models/cosmos3/
+  - vllm_omni/diffusion/distributed/autoencoders/autoencoder_kl_wan.py
+  - vllm_omni/experimental/world_models/
+  - vllm_omni/model_extras/cosmos3.py
+
+x-qwen-image: &qwen_image
+  - vllm_omni/diffusion/models/qwen_image/
+  - vllm_omni/diffusion/models/dmd2/
+  - vllm_omni/diffusion/distributed/autoencoders/autoencoder_kl_qwenimage.py
+
+x-joy-image: &joy_image
+  - vllm_omni/diffusion/models/joy_image/
+  - vllm_omni/diffusion/distributed/autoencoders/autoencoder_kl_wan.py
+
+x-hunyuan-image3: &hunyuan_image3
+  - vllm_omni/diffusion/models/hunyuan_image3/
+  - vllm_omni/model_executor/models/hunyuan_image3/
+  - vllm_omni/deploy/hunyuan_image3_dit.yaml
+
+x-minimax-h3: &minimax_h3
+  - vllm_omni/diffusion/models/minimax_h3/
+  - vllm_omni/diffusion/layers/indexed_modulation.py
+
+x-hunyuan-video: &hunyuan_video
+  - vllm_omni/diffusion/models/hunyuan_video/
+  - vllm_omni/model_executor/models/hunyuan_video/
+
+x-ltx2: &ltx2
+  - vllm_omni/diffusion/models/ltx2/
+  - vllm_omni/model_extras/ltx2.py
+
+x-sana-video: &sana_video
+  - vllm_omni/diffusion/models/sana_video/
+  - vllm_omni/model_extras/sana_video.py
+
+x-z-image: &z_image
+  - vllm_omni/diffusion/models/z_image/
+
+x-stable-audio: &stable_audio
+  - vllm_omni/diffusion/models/stable_audio/
+
+x-lingbot-video: &lingbot_video
+  - vllm_omni/diffusion/models/lingbot_video/
+  - vllm_omni/model_extras/lingbot_video.py
+
+x-lingbot-world: &lingbot_world
+  - vllm_omni/diffusion/models/lingbot_world/
+
+x-kandinsky6: &kandinsky6
+  - vllm_omni/diffusion/models/kandinsky6/
+  - vllm_omni/model_extras/kandinsky6.py
+
+
+source_file_dependencies:
+  # Hardware presets and shared test helpers. Not a job key — do not attach this
+  # to a step. Used only when no job-key prefix matches: then a hit keeps every job.
+  source_filter_fallback:
+    - .buildkite/common/ci_mirror_hardwares.yml
+    - tests/helpers/
+
+  # --- features ---
+  diffusion_distributed_attention:
+    - tests/diffusion/distributed/
+    - tests/diffusion/attention/test_attention_sp.py
+    - vllm_omni/diffusion/distributed/
+    - vllm_omni/diffusion/attention/
+    - vllm_omni/diffusion/data.py
+    - vllm_omni/diffusion/models/
+
+  custom_pipeline:
+    - tests/e2e/features/custom_pipeline/
+    - tests/e2e/features/helpers/custom_pipeline.py
+    - vllm_omni/diffusion/
+    - vllm_omni/engine/
+    - vllm_omni/entrypoints/
+    - vllm_omni/inputs/
+    - vllm_omni/outputs/
+
+  rlhf_verl_omni:
+    - tests/e2e/features/rlhf_test/test_verl_omni_e2e.py
+    - tests/e2e/features/helpers/verl_omni_server.py
+    - tests/e2e/features/helpers/custom_pipeline.py
+    - vllm_omni/diffusion/
+    - vllm_omni/engine/
+    - vllm_omni/entrypoints/
+    - vllm_omni/inputs/
+    - vllm_omni/outputs/
+
+  # Tiny-model diffusion jobs (shared across many model trees under models/).
+  diffusion_tiny_model:
+    - tests/model_tests/diffusion/
+    - vllm_omni/diffusion/models/
+
+  diffusion_offloader:
+    - tests/diffusion/offloader/
+    - vllm_omni/diffusion/offloader/
+
+  diffusion_quantization:
+    - tests/diffusion/quantization/
+    - vllm_omni/diffusion/quantization/
+
+  # --- models ---
+  # --- omni / tts jobs ---
+  entrypoints_function:
+    - tests/entrypoints/
+    - vllm_omni/entrypoints/
+
+  omni_qwen3_omni_function:
+    - *qwen3_omni
+    - tests/e2e/offline_inference/test_qwen3_omni.py
+    - tests/e2e/offline_inference/test_qwen3_omni_colocate_async.py
+    - tests/e2e/offline_inference/test_qwen3_omni_autoround_w4a16_expansion.py
+    - tests/e2e/offline_inference/test_qwen3_omni_modelopt_nvfp4_w4a4.py
+    - tests/e2e/online_serving/test_qwen3_omni.py
+    - tests/e2e/online_serving/test_qwen3_omni_expansion.py
+    - tests/e2e/online_serving/test_qwen3_omni_multi_replicas.py
+
+  omni_qwen3_omni_accuracy:
+    - *qwen3_omni
+    - tests/e2e/accuracy/qwen3_omni/
+
+  omni_qwen3_omni_perf:
+    - *qwen3_omni
+    - tests/dfx/perf/scripts/run_benchmark.py
+    - tests/dfx/perf/tests/test_qwen3_omni_vllm_text.json
+    - tests/dfx/perf/tests/test_qwen3_omni_async_chunk.json
+    - tests/dfx/perf/tests/test_qwen3_omni_no_async_chunk.json
+    - tests/dfx/perf/tests/test_qwen3_omni_multi_replicas.json
+
+  omni_qwen3_omni_reliability:
+    - *qwen3_omni
+    - tests/dfx/reliability/test_reliability_qwen3_omni.py
+
+  omni_qwen3_omni_doc:
+    - *qwen3_omni
+    - examples/online_serving/qwen3_omni/
+    - examples/online_serving/openai_chat_completion_client_for_multimodal_generation.py
+    - tests/examples/online_serving/test_qwen3_omni.py
+
+  omni_qwen2_5_omni_doc:
+    - *qwen2_5_omni
+    - examples/online_serving/qwen2_5_omni/
+    - examples/online_serving/openai_chat_completion_client_for_multimodal_generation.py
+    - tests/examples/online_serving/test_qwen2_5_omni.py
+
+  omni_minicpmo_4_5_function:
+    - *minicpmo_4_5
+    - tests/e2e/offline_inference/test_minicpmo_4_5.py
+    - tests/e2e/online_serving/test_minicpmo_4_5.py
+    - tests/e2e/online_serving/test_minicpmo_4_5_expansion.py
+
+  omni_minicpmo_4_5_accuracy:
+    - *minicpmo_4_5
+    - tests/e2e/accuracy/minicpmo_4_5/
+
+  omni_duplex_eval_ci:
+    - *minicpmo_4_5
+    - *minicpmo_4_5_duplex
+    - tests/e2e/accuracy/omni_duplex_eval/**
+    - tests/e2e/online_serving/helpers/minicpmo_4_5_duplex.py
+    - vllm_omni/benchmarks/duplex/**
+    - vllm_omni/entrypoints/cli/benchmark/omni_duplex_eval.py
+
+  omni_minicpmo_4_5_perf:
+    - *minicpmo_4_5
+    - tests/dfx/perf/scripts/run_benchmark.py
+    - tests/dfx/perf/tests/test_minicpmo_4_5_ready.json
+    - tests/dfx/perf/tests/test_minicpmo_4_5.json
+
+  omni_minicpmo_4_5_doc:
+    - *minicpmo_4_5
+    - examples/online_serving/minicpmo/realtime_web/
+    - examples/online_serving/minicpmo/realtime_duplex_demo.py
+    - tests/examples/test_minicpmo_realtime_web_server.py
+    - tests/examples/test_minicpmo_realtime_web_static.py
+    - tests/examples/test_minicpmo_realtime_duplex_simple_demo.py
+
+  omni_minicpmo_4_5_duplex_function:
+    - *minicpmo_4_5
+    - *minicpmo_4_5_duplex
+    - tests/e2e/online_serving/test_minicpmo_4_5_duplex.py
+    - tests/e2e/online_serving/test_minicpmo_4_5_duplex_expansion.py
+    - tests/e2e/online_serving/test_minicpmo_4_5_window.py
+    - tests/e2e/online_serving/helpers/minicpmo_window_e2e.py
+    - tests/e2e/online_serving/test_duplex_client_live.py
+    - tests/e2e/online_serving/test_minicpmo_realtime_duplex_drivers.py
+    - tests/e2e/online_serving/helpers/minicpmo_realtime_duplex_scenarios.py
+    - tests/e2e/online_serving/run_minicpmo_realtime_duplex_multi_session.py
+    - tests/e2e/online_serving/run_minicpmo_realtime_duplex_server_vad.py
+    - tests/e2e/online_serving/run_minicpmo_realtime_duplex_soft_interrupt.py
+    - tests/e2e/online_serving/helpers/minicpmo_4_5_duplex.py
+    - tests/assets/minicpmo_4_5/
+    - tests/clients/
+    - tests/engine/duplex/
+    - tests/engine/test_duplex_omni_engine.py
+    - tests/engine/test_duplex_orchestrator.py
+    - tests/entrypoints/duplex/
+    - tests/entrypoints/test_duplex_omni.py
+    - tests/entrypoints/openai_api/test_duplex_api_server.py
+    - tests/model_executor/models/minicpmo_4_5/duplex/
+    - tests/protocol/
+
+  omni_minicpmo_4_5_duplex_perf:
+    - *minicpmo_4_5
+    - *minicpmo_4_5_duplex
+    - tests/dfx/perf/scripts/run_benchmark.py
+    - tests/dfx/perf/tests/test_minicpmo_4_5_duplex_seed_tts.json
+
+  tts_qwen3_tts_function:
+    - *qwen3_tts
+    - tests/e2e/features/prefill_phase/
+    - tests/e2e/offline_inference/test_qwen3_tts_base.py
+    - tests/e2e/offline_inference/test_qwen3_tts_customvoice.py
+    - tests/e2e/online_serving/test_qwen3_tts_base.py
+    - tests/e2e/online_serving/test_qwen3_tts_base_expansion.py
+    - tests/e2e/online_serving/test_qwen3_tts_customvoice.py
+    - tests/e2e/online_serving/test_qwen3_tts_customvoice_expansion.py
+
+  tts_qwen3_tts_perf:
+    - *qwen3_tts
+    - tests/dfx/perf/scripts/run_benchmark.py
+    - tests/dfx/perf/tests/test_tts.json
+
+  tts_qwen3_tts_cov:
+    - *qwen3_tts
+    - .buildkite/common/scripts/run_cov_split.sh
+    - pyproject.toml
+    - tests/e2e/offline_inference/test_qwen3_tts_base.py
+    - tests/e2e/online_serving/test_qwen3_tts_base.py
+
+  # --- diffusion jobs ---
+  diffusion_wan22_function:
+    - *wan22
+    - tests/e2e/offline_inference/test_wan22_t2v.py
+    - tests/e2e/offline_inference/test_wan22_autoround_w4a16_expansion.py
+    - tests/e2e/offline_inference/test_device_postprocess_equivalence.py
+    - tests/e2e/online_serving/test_wan22_t2v.py
+    - tests/e2e/online_serving/test_wan22_expansion.py
+
+  diffusion_wan22_accuracy:
+    - *wan22
+    - tests/e2e/accuracy/wan22_i2v/
+    - tests/e2e/accuracy/test_diffusers_backend_similarity.py
+
+  diffusion_wan22_perf:
+    - *wan22
+    - tests/dfx/perf/scripts/run_benchmark.py
+    - tests/dfx/perf/tests/test_wan22_i2v_vllm_omni.json
+
+  diffusion_wan22_reliability:
+    - *wan22
+    - tests/dfx/reliability/test_reliability_wan22.py
+
+  diffusion_image_to_video_doc:
+    - *wan22
+    - *hunyuan_video
+    - *ltx2
+    - *sana_video
+    - *lingbot_video
+    - examples/offline_inference/image_to_video/
+    - tests/examples/offline_inference/test_image_to_video.py
+
+  diffusion_cosmos3_function:
+    - *cosmos3
+    - tests/e2e/online_serving/test_cosmos3.py
+
+  diffusion_cosmos3_accuracy:
+    - *cosmos3
+    - tests/e2e/accuracy/test_cosmos3_similarity.py
+
+  diffusion_cosmos3_perf:
+    - *cosmos3
+    - tests/dfx/perf/scripts/run_benchmark.py
+    - tests/dfx/perf/tests/test_cosmos3_vllm_omni.json
+
+  diffusion_qwen_image_function:
+    - *qwen_image
+    - tests/e2e/online_serving/test_qwen_image.py
+    - tests/e2e/online_serving/test_qwen_image_expansion.py
+    - tests/e2e/offline_inference/test_qwen_image_autoround_w4a16_expansion.py
+
+  diffusion_qwen_image_accuracy:
+    - *qwen_image
+    - tests/e2e/accuracy/test_qwen_image.py
+    - tests/e2e/accuracy/test_diffusers_backend_similarity.py
+
+  diffusion_joy_image_accuracy:
+    - *joy_image
+    - tests/diffusion/models/joy_image/
+    - tests/e2e/accuracy/test_joy_image_edit.py
+    - recipes/JD/JoyAI-Image-Edit.md
+
+  diffusion_qwen_image_perf:
+    - *qwen_image
+    - tests/dfx/perf/scripts/run_diffusion_benchmark.py
+    - tests/dfx/perf/tests/test_qwen_image_vllm_omni.json
+
+  diffusion_text_to_image_doc:
+    - *qwen_image
+    - *z_image
+    - examples/offline_inference/text_to_image/
+    - examples/online_serving/text_to_image/
+    - tests/examples/offline_inference/test_text_to_image.py
+    - tests/examples/online_serving/test_text_to_image.py
+    - vllm_omni/diffusion/models/nextstep_1_1/
+    - vllm_omni/diffusion/models/longcat_image/
+    - vllm_omni/diffusion/models/ovis_image/
+    - vllm_omni/diffusion/models/omnigen2/
+    - vllm_omni/diffusion/models/sd3/
+    - vllm_omni/diffusion/models/flux/
+    - vllm_omni/diffusion/models/flux2/
+    - vllm_omni/diffusion/models/flux2_klein/
+    - vllm_omni/diffusion/models/hidream_image/
+    - vllm_omni/diffusion/models/krea2/
+    - vllm_omni/diffusion/models/lingbot_video/
+
+  diffusion_hunyuan_image3_function:
+    - *hunyuan_image3
+    - tests/e2e/offline_inference/test_hunyuanimage3_i2t_expansion.py
+
+  diffusion_hunyuan_image3_accuracy:
+    - *hunyuan_image3
+    - tests/e2e/accuracy/test_hunyuan_image3_pixel_accuracy.py
+    - tests/e2e/accuracy/test_hunyuan_image3_lora.py
+    - vllm_omni/diffusion/lora/
+
+  diffusion_hunyuan_image3_perf:
+    - *hunyuan_image3
+    - tests/dfx/perf/scripts/run_diffusion_benchmark.py
+    - tests/dfx/perf/tests/test_hunyuan_image_tp2_cfgp2.json
+    - tests/dfx/perf/tests/test_hunyuan_image_tp2_sp2.json
+    - tests/dfx/perf/tests/test_hunyuan_image_tp4.json
+
+  diffusion_hunyuan_image3_reliability:
+    - *hunyuan_image3
+    - tests/dfx/reliability/test_reliability_hunyuan_image.py
+
+  diffusion_minimax_h3_function:
+    - *minimax_h3
+    - tests/helpers/assertions.py
+    - tests/e2e/online_serving/test_minimax_h3_dlo_dp2_t2va.py
+    - tests/e2e/online_serving/minimax_h3/
+    - tests/diffusion/models/minimax_h3/
+    - vllm_omni/diffusion/lora/
+    - vllm_omni/diffusion/offloader/
+    - vllm_omni/diffusion/distributed/
+    - vllm_omni/entrypoints/openai/
+
+  diffusion_minimax_h3_accuracy:
+    - *minimax_h3
+    - tests/e2e/accuracy/minimax_h3/
+
+  diffusion_minimax_h3_perf:
+    - *minimax_h3
+    - tests/dfx/perf/scripts/run_benchmark.py
+    - tests/dfx/perf/tests/test_minimax_h3_vllm_omni.json
+
+  diffusion_hunyuan_video_function:
+    - *hunyuan_video
+    - tests/e2e/online_serving/test_hunyuan_video_15_expansion.py
+
+  diffusion_hunyuan_video_accuracy:
+    - *hunyuan_video
+    - tests/e2e/accuracy/hunyuanvideo15_t2v/
+    - tests/e2e/accuracy/hunyuanvideo15_i2v/
+
+  diffusion_hunyuan_video_perf:
+    - *hunyuan_video
+    - tests/dfx/perf/scripts/run_benchmark.py
+    - tests/dfx/perf/tests/test_hunyuanvideo15_t2v_vllm_omni.json
+    - tests/dfx/perf/tests/test_hunyuanvideo15_i2v_vllm_omni.json
+
+  diffusion_z_image_function:
+    - *z_image
+    - tests/e2e/online_serving/test_z_image.py
+    - tests/e2e/online_serving/test_zimage_expansion.py
+
+  diffusion_sana_video_function:
+    - *sana_video
+    - tests/e2e/online_serving/test_sana_video_expansion.py
+    - tests/e2e/accuracy/sana_video/
+
+  diffusion_stable_audio_function:
+    - *stable_audio
+    - tests/e2e/offline_inference/test_stable_audio_expansion.py
+    - tests/e2e/online_serving/test_stable_audio_online_expansion.py
+
+  diffusion_lingbot_function:
+    - *lingbot_video
+    - *lingbot_world
+    - tests/e2e/online_serving/test_lingbot_video.py
+    - tests/e2e/online_serving/test_lingbot_video_expansion.py
+    - tests/e2e/online_serving/test_lingbot_video_moe.py
+    - tests/e2e/offline_inference/test_lingbot_world_v2.py
+
+  diffusion_lingbot_perf:
+    - *lingbot_video
+    - tests/dfx/perf/scripts/run_benchmark.py
+    - tests/dfx/perf/tests/test_lingbot_video_vllm_omni.json
+
+  diffusion_kandinsky6_function:
+    - *kandinsky6
+    - tests/e2e/offline_inference/test_kandinsky6.py
+    - tests/e2e/online_serving/test_kandinsky6.py
+
+
+  # Invalid-param jobs: model business code only, plus the reliability scripts.
+  reliability_invalid_param_h100:
+    - *qwen3_omni
+    - *minicpmo_4_5
+    - *qwen_image
+    - *wan22
+    - tests/dfx/reliability/invalid_param_test/
+
+  reliability_invalid_param_l4:
+    - *qwen3_tts
+    - *stable_audio
+    - tests/dfx/reliability/invalid_param_test/

.buildkite/common/scripts/skip_ci.py

@@ -44,6 +44,7 @@ L3_YAML_FILES: dict[str, str] = {
 L45_YAML_FILES: dict[str, str] = {
     ".buildkite/cuda/test-nightly.yml": "cuda",
     ".buildkite/cuda/test-weekly.yml": "cuda",
+    ".buildkite/amd/test-amd-nightly.yml": "amd",
     ".buildkite/npu/test-npu-nightly.yml": "npu",
 }
 

.buildkite/common/scripts/upload_pipeline.py

@@ -10,8 +10,17 @@ Bootstrap mode (``bootstrap-upload-steps.yml``):
   - Detect docs-only, pytest skip-mark-only, or combined skip-ci from git diff.
   - When only CI level YAML changes, enable **L2/L3** upload steps for affected levels only.
 
-Test pipeline mode (e.g. test-merge.yml):
-  - Drop steps whose ``source_file_dependencies`` do not match changed files.
+Test pipeline mode (e.g. test-merge.yml, test-nightly.yml, test-weekly.yml):
+  - Drop steps whose ``source_file_dependencies`` do not match changed files
+    (string/list keys from ci_source_file_dependencies.yml, or inline path
+    prefixes). Filtering applies on **PR label** uploads and on post-merge
+    ``main`` L3 uploads. Scheduled ``main`` uploads that pass ``--all``
+    (NIGHTLY/WEEKLY) or ``--e2e`` keep every selected job and still strip
+    the field. If no job-key prefix matches, a change to the pipeline YAML
+    being uploaded or to a path under the ``source_filter_fallback`` registry
+    key keeps every job so command, env, and hardware edits can be validated
+    before merge. When any job-key prefix already matches, normal filtering
+    wins.
   - Expand uploader-only ``mirror_hardwares`` into ``agents`` (+ optional ``image``
     for NPU) + ``plugins`` (see ci_mirror_hardwares.yml).
   - Omit ``mirror_hardwares`` to compose ``{chip}_{n}`` from pytest ``-m`` SKU
@@ -87,22 +96,20 @@ BOOTSTRAP_UPLOAD_IF_KEYS = {
 }
 E2E_GROUP_MARKER = "E2E Test"
 CI_MIRROR_HARDWARES_PATH = ROOT / ".buildkite/common/ci_mirror_hardwares.yml"
+CI_SOURCE_FILE_DEPENDENCIES_PATH = ROOT / ".buildkite/common/ci_source_file_dependencies.yml"
+# Registry key whose prefixes bypass source filtering. Paths live in
+# ci_source_file_dependencies.yml; do not attach this key to a job.
+SOURCE_FILTER_FALLBACK_KEY = "source_filter_fallback"
 CUDA_HF_TOKEN_ENV = "VLLM_CI_HF_TOKEN"
 CUDA_HF_TOKEN_EXPORT = f'if [ -n "$${{{CUDA_HF_TOKEN_ENV}:-}}" ]; then export HF_TOKEN="$${{{CUDA_HF_TOKEN_ENV}}}"; fi'
+CUDA_RUNTIME_CHECK = "python3 .buildkite/cuda/scripts/check_vllm_runtime.py"
 
 # Bootstrap Buildkite ``if`` expressions.
 # ``*_MAIN_IF``: main + env schedule. ``*_LABEL_IF``: PR label (and/or composed with MAIN).
 # ``*_UPLOAD_IF``: full gate for uploading that child pipeline.
 NIGHTLY_MAIN_IF = 'build.branch == "main" && build.env("NIGHTLY") == "1"'
 NIGHTLY_LABEL_IF = (
-    f"({NIGHTLY_MAIN_IF}) || "
-    '(build.branch != "main" && ('
-    'build.pull_request.labels includes "nightly-test" || '
-    'build.pull_request.labels includes "omni-test" || '
-    'build.pull_request.labels includes "tts-test" || '
-    'build.pull_request.labels includes "diffusion-x2iat-test" || '
-    'build.pull_request.labels includes "diffusion-x2v-test"'
-    "))"
+    f'({NIGHTLY_MAIN_IF}) || (build.branch != "main" && build.pull_request.labels includes "nightly-test")'
 )
 WEEKLY_E2E_IF = 'build.branch == "main" && build.env("WEEKLY") == "1"'
 WEEKLY_MAIN_IF = 'build.branch == "main" && (build.env("WEEKLY") == "1" || build.env("NON_CRITICAL") == "1")'
@@ -248,7 +255,7 @@ def _render_bootstrap_pipeline(
     return yaml.safe_dump(doc, sort_keys=False)
 
 
-# --- Test pipeline (test-ready.yml, test-merge.yml) ---
+# --- Test pipeline (test-ready.yml, test-merge.yml, test-nightly.yml) ---
 
 
 @lru_cache(maxsize=1)
@@ -264,6 +271,117 @@ def _load_mirror_hardwares() -> dict[str, dict[str, Any]]:
     return presets
 
 
+@lru_cache(maxsize=1)
+def _load_source_file_dependencies() -> dict[str, list[str]]:
+    def flatten(value: Any, *, key: str) -> list[str]:
+        """Flatten YAML-anchor nested lists into a de-duplicated prefix list."""
+        if isinstance(value, str):
+            if not value.strip():
+                raise ValueError(f"empty path in source_file_dependencies[{key!r}]")
+            return [value]
+        if isinstance(value, list):
+            flattened: list[str] = []
+            seen: set[str] = set()
+            for item in value:
+                for path in flatten(item, key=key):
+                    if path not in seen:
+                        seen.add(path)
+                        flattened.append(path)
+            return flattened
+        raise ValueError(
+            f"source_file_dependencies[{key!r}] must be a list of path prefixes, got {type(value).__name__}",
+        )
+
+    if not CI_SOURCE_FILE_DEPENDENCIES_PATH.is_file():
+        raise FileNotFoundError(
+            f"missing CI source_file_dependencies registry: {CI_SOURCE_FILE_DEPENDENCIES_PATH}",
+        )
+    doc = yaml.safe_load(CI_SOURCE_FILE_DEPENDENCIES_PATH.read_text(encoding="utf-8"))
+    if not isinstance(doc, dict):
+        raise ValueError(f"invalid CI source_file_dependencies registry: {CI_SOURCE_FILE_DEPENDENCIES_PATH}")
+    presets = doc.get("source_file_dependencies")
+    if not isinstance(presets, dict):
+        raise ValueError(f"source_file_dependencies must be a mapping in {CI_SOURCE_FILE_DEPENDENCIES_PATH}")
+    loaded: dict[str, list[str]] = {}
+    for key, value in presets.items():
+        if not isinstance(key, str) or not key.strip():
+            raise ValueError(
+                f"source_file_dependencies keys must be non-empty strings in {CI_SOURCE_FILE_DEPENDENCIES_PATH}",
+            )
+        loaded[key] = flatten(value, key=key)
+    return loaded
+
+
+_REGISTRY_KEY_RE = re.compile(r"^[a-z][a-z0-9_]*$")
+
+
+def _parse_command_text(step: dict[str, Any]) -> str:
+    """Parse step ``commands`` / ``command`` into a single string."""
+    raw = step.get("commands", step.get("command"))
+    if raw is None:
+        return ""
+    if isinstance(raw, str):
+        return raw
+    if isinstance(raw, list):
+        return "\n".join(str(item) for item in raw if item is not None)
+    return str(raw)
+
+
+def _resolve_source_file_dependencies(step: dict[str, Any]) -> list[str] | None:
+    """Expand a registry key (or list of keys / inline path prefixes)."""
+    deps = step.get("source_file_dependencies")
+    if deps is None:
+        return None
+
+    def dedupe(*groups: list[str]) -> list[str]:
+        merged: list[str] = []
+        seen: set[str] = set()
+        for group in groups:
+            for path in group:
+                if path not in seen:
+                    seen.add(path)
+                    merged.append(path)
+        return merged
+
+    def lookup(key: str) -> list[str]:
+        if key == SOURCE_FILTER_FALLBACK_KEY:
+            raise ValueError(
+                f"source_file_dependencies {key!r} bypasses filtering and cannot be attached to a step "
+                f"({_get_step_label(step)!r})",
+            )
+        registry = _load_source_file_dependencies()
+        paths = registry.get(key)
+        if paths is None:
+            known = ", ".join(sorted(registry))
+            raise ValueError(
+                f"unknown source_file_dependencies {key!r} in step {_get_step_label(step)!r}; known: {known}",
+            )
+        return list(paths)
+
+    prefixes: list[str]
+    if isinstance(deps, str):
+        prefixes = lookup(deps) if _REGISTRY_KEY_RE.fullmatch(deps) else [deps]
+    elif isinstance(deps, list):
+        if not all(isinstance(item, str) for item in deps):
+            raise ValueError(
+                f"source_file_dependencies must be a string key or list of strings in step {_get_step_label(step)!r}",
+            )
+        keyish = [bool(_REGISTRY_KEY_RE.fullmatch(item)) for item in deps]
+        if deps and all(keyish):
+            prefixes = dedupe(*(lookup(key) for key in deps))
+        elif any(keyish):
+            raise ValueError(
+                f"source_file_dependencies in step {_get_step_label(step)!r} mixes registry keys and path prefixes",
+            )
+        else:
+            prefixes = list(deps)
+    else:
+        raise ValueError(
+            f"source_file_dependencies must be a string key or list in step {_get_step_label(step)!r}",
+        )
+    return prefixes
+
+
 @lru_cache(maxsize=1)
 def _cuda_mirror_chips() -> tuple[str, ...]:
     """Lowercase CUDA SKUs that have at least one ``{chip}_{n}`` preset.
@@ -278,7 +396,7 @@ def _cuda_mirror_chips() -> tuple[str, ...]:
     return tuple(sorted(chips, key=lambda chip: (-len(chip), chip)))
 
 
-def _read_cards_marks(expr: str) -> tuple[set[int], bool]:
+def _parse_cards_marks(expr: str) -> tuple[set[int], bool]:
     """Parse registered ``cards_n`` / ``not cards_n`` from a pytest ``-m`` expr.
 
     Returns positive card counts and whether any ``not cards_*`` is present.
@@ -295,8 +413,8 @@ def _read_cards_marks(expr: str) -> tuple[set[int], bool]:
     return positives, has_not_cards
 
 
-def _read_hardware_marks(expr: str) -> set[str]:
-    """Return lowercase positive CUDA SKUs that have a mirror preset.
+def _parse_hardware_marks(expr: str) -> set[str]:
+    """Parse lowercase positive CUDA SKUs that have a mirror preset.
 
     ``not H100`` is ignored. SKUs with no ``{chip}_*`` preset (e.g. H200) are dropped.
     """
@@ -338,36 +456,22 @@ def _get_mirror_hw_selector() -> str:
     return selector
 
 
-def _read_pytest_marks(commands: Any) -> tuple[set[str], int | Literal["max"] | None]:
-    """Read pytest ``-m`` from step commands: CUDA SKU chips and cards count.
+def _parse_pytest_marks(step: dict[str, Any]) -> tuple[set[str], int | Literal["max"] | None]:
+    """Parse pytest ``-m`` from step commands: CUDA SKU chips and cards count.
 
     *chips* are lowercase SKUs from positive markers (``not H100`` is ignored).
     *cards* is the max positive ``cards_n``, ``"max"`` when only ``not cards_*``
     is present, or ``None``.
     """
-    chunks: list[str] = []
-
-    def _collect(value: Any) -> None:
-        if value is None:
-            return
-        if isinstance(value, str):
-            chunks.append(value)
-        elif isinstance(value, list):
-            for part in value:
-                _collect(part)
-        else:
-            chunks.append(str(value))
-
-    _collect(commands)
-    text = "\n".join(chunks)
+    text = _parse_command_text(step)
 
     chips: set[str] = set()
     positives: set[int] = set()
     has_not_cards = False
     for match in _PYTEST_MARKER_ARG.finditer(text):
         expr = match.group(1) or match.group(2) or match.group(3) or ""
-        chips |= _read_hardware_marks(expr)
-        card_counts, not_cards = _read_cards_marks(expr)
+        chips |= _parse_hardware_marks(expr)
+        card_counts, not_cards = _parse_cards_marks(expr)
         positives.update(card_counts)
         has_not_cards = has_not_cards or not_cards
 
@@ -484,7 +588,7 @@ def _expand_mirror_hardwares(step: dict[str, Any]) -> dict[str, Any] | None:
     if "mirror_hardwares" not in step:
         if has_pool:
             return step
-        chips, cards = _read_pytest_marks(step.get("commands"))
+        chips, cards = _parse_pytest_marks(step)
         if not chips and cards is None:
             return step
         preset_name = _compose_mirror_hardware_name(chips, cards, step_label=step_label)
@@ -517,11 +621,11 @@ def _expand_mirror_hardwares(step: dict[str, Any]) -> dict[str, Any] | None:
     if any(preset_name == chip or preset_name.startswith(f"{chip}_") for chip in _cuda_mirror_chips()):
         commands = merged.get("commands")
         if isinstance(commands, list):
-            merged["commands"] = [CUDA_HF_TOKEN_EXPORT, *commands]
+            merged["commands"] = [CUDA_HF_TOKEN_EXPORT, CUDA_RUNTIME_CHECK, *commands]
         elif commands is None:
-            merged["commands"] = [CUDA_HF_TOKEN_EXPORT]
+            merged["commands"] = [CUDA_HF_TOKEN_EXPORT, CUDA_RUNTIME_CHECK]
         else:
-            merged["commands"] = [CUDA_HF_TOKEN_EXPORT, commands]
+            merged["commands"] = [CUDA_HF_TOKEN_EXPORT, CUDA_RUNTIME_CHECK, commands]
     # Preset retry (K8S_RETRY on l4_*) must not clobber a step that opted out.
     if "retry" in step:
         merged["retry"] = step["retry"]
@@ -552,11 +656,7 @@ def _process_test_steps(
             processed.append(step)
             continue
 
-        deps = step.get("source_file_dependencies")
-        if deps is not None and not isinstance(deps, list):
-            raise ValueError(
-                f"source_file_dependencies must be a list in step {_get_step_label(step)!r}",
-            )
+        deps = _resolve_source_file_dependencies(step)
         if changed_files is not None and deps is not None and not _match_source_file(changed_files, deps):
             _log(f"skip {_get_step_label(step)!r} (no changes under {deps})")
             continue
@@ -595,10 +695,25 @@ def _select_e2e_group_steps(steps: list[Any]) -> list[Any]:
     return selected
 
 
+def _any_source_dependency_match(steps: list[Any], changed_files: list[str]) -> bool:
+    """True when any step with ``source_file_dependencies`` matches *changed_files*."""
+    for step in steps:
+        if not isinstance(step, dict):
+            continue
+        deps = _resolve_source_file_dependencies(step)
+        if deps is not None and _match_source_file(changed_files, deps):
+            return True
+        nested = step.get("steps")
+        if isinstance(nested, list) and _any_source_dependency_match(nested, changed_files):
+            return True
+    return False
+
+
 def _render_test_pipeline(
     doc: dict[str, Any],
     changed_files: list[str] | None,
     *,
+    pipeline_path: Path | None = None,
     e2e_only: bool = False,
 ) -> dict[str, Any]:
     """Filter steps by PR diff and strip uploader-only ``source_file_dependencies`` metadata."""
@@ -611,6 +726,13 @@ def _render_test_pipeline(
         return doc
     if e2e_only:
         steps = _select_e2e_group_steps(steps)
+    # Bypass is a fallback: only when no job-key prefix matched.
+    if changed_files is not None and pipeline_path is not None:
+        if not _any_source_dependency_match(steps, changed_files):
+            bypass = _source_filter_fallback_reason(changed_files, pipeline_path)
+            if bypass is not None:
+                _log(f"keep all jobs (no source_file_dependencies match; bypassed by {bypass})")
+                changed_files = None
     steps = _process_test_steps(steps, changed_files)
     return {**doc, "steps": steps}
 
@@ -618,6 +740,60 @@ def _render_test_pipeline(
 # --- Entry (read file → bootstrap or test render → YAML string) ---
 
 
+def _repo_relative_posix(path: Path) -> str:
+    resolved = path.resolve()
+    try:
+        return resolved.relative_to(ROOT).as_posix()
+    except ValueError:
+        return path.as_posix().replace("\\", "/")
+
+
+def _source_filter_fallback_prefixes() -> list[str]:
+    """Shared uploader/registry prefixes from ``source_filter_fallback``. Not a job key."""
+    registry = _load_source_file_dependencies()
+    prefixes = registry.get(SOURCE_FILTER_FALLBACK_KEY)
+    if not prefixes:
+        raise ValueError(
+            f"source_file_dependencies[{SOURCE_FILTER_FALLBACK_KEY!r}] must list shared uploader paths "
+            f"in {CI_SOURCE_FILE_DEPENDENCIES_PATH}",
+        )
+    return prefixes
+
+
+def _source_filter_fallback_reason(changed_files: list[str] | None, pipeline_path: Path) -> str | None:
+    """Return the changed path that should disable source filtering, if any."""
+    if not changed_files:
+        return None
+    changed = set(changed_files)
+    pipeline_rel = _repo_relative_posix(pipeline_path)
+    if pipeline_rel in changed:
+        return pipeline_rel
+    prefixes = _source_filter_fallback_prefixes()
+    for path in changed_files:
+        if _match_source_file([path], prefixes):
+            return path
+    return None
+
+
+def _changed_files_for_source_filter(
+    ctx,
+    *,
+    force_all: bool,
+    e2e_only: bool,
+) -> list[str] | None:
+    """Return the diff to filter against, or None to keep every step.
+
+    ``--all`` / ``--e2e`` disable filtering (scheduled NIGHTLY/WEEKLY full
+    uploads and weekly E2E sweeps). Post-merge ``main`` L3 and PR-label
+    uploads filter against the commit/PR diff. Pipeline YAML /
+    ``source_filter_fallback`` matches are applied later in
+    ``_render_test_pipeline`` only when no job-key prefix already matched.
+    """
+    if force_all or e2e_only:
+        return None
+    return ctx.changed_files
+
+
 def _render_pipeline(
     path: Path,
     *,
@@ -635,13 +811,18 @@ def _render_pipeline(
 
     text = path.read_text(encoding="utf-8")
     ctx = resolve_ci_context_from_git()
-    changed_files = None if force_all or e2e_only else ctx.changed_files
+    changed_files = _changed_files_for_source_filter(ctx, force_all=force_all, e2e_only=e2e_only)
 
     doc = yaml.safe_load(text)
     if not isinstance(doc, dict):
         raise ValueError(f"invalid pipeline YAML: {path}")
 
-    doc = _render_test_pipeline(doc, changed_files, e2e_only=e2e_only)
+    doc = _render_test_pipeline(
+        doc,
+        changed_files,
+        pipeline_path=path,
+        e2e_only=e2e_only,
+    )
     return yaml.safe_dump(doc, sort_keys=False)
 
 

.buildkite/cuda/bootstrap-upload-steps.yml

@@ -22,7 +22,8 @@ steps:
     agents:
       queue: "cpu_queue_premerge"
 
-  # L3 Test — PR merge-test label (diff-aware) or main+WEEKLY=1 (--e2e, full E2E group)
+  # L3 Test — PR merge-test label or post-merge on main (both diff-aware);
+  # main+WEEKLY=1 uploads the E2E group only (--e2e).
   - label: "Upload Merge Pipeline"
     depends_on: image-build
     key: upload-merge-pipeline
@@ -36,20 +37,30 @@ steps:
     agents:
       queue: "cpu_queue_premerge"
 
-  # L4 Test — main+NIGHTLY=1 (scheduled), or PR with specific label (e.g. add label then Rebuild)
+  # L4 Test — main+NIGHTLY=1 (full pipeline), or PR nightly-test label (diff-aware)
   - label: "Upload Nightly Pipeline"
     depends_on: image-build
     key: upload-nightly-pipeline
     commands:
-      - python3 .buildkite/common/scripts/upload_pipeline.py --upload .buildkite/cuda/test-nightly.yml
+      - |
+        EXTRA=""
+        if [ "${NIGHTLY:-}" = "1" ] && [ "${BUILDKITE_BRANCH}" = "main" ]; then
+          EXTRA="--all"
+        fi
+        python3 .buildkite/common/scripts/upload_pipeline.py --upload $$EXTRA .buildkite/cuda/test-nightly.yml
     agents:
       queue: "cpu_queue_premerge"
 
-  # L5 Test — main+WEEKLY=1 / NON_CRITICAL=1 (scheduled), or PR with weekly-test label
+  # L5 Test — main+WEEKLY=1 / NON_CRITICAL=1 (full pipeline), or PR weekly-test label (diff-aware)
   - label: "Upload Weekly Pipeline"
     depends_on: image-build
     key: upload-weekly-pipeline
     commands:
-      - python3 .buildkite/common/scripts/upload_pipeline.py --upload .buildkite/cuda/test-weekly.yml
+      - |
+        EXTRA=""
+        if [ "${BUILDKITE_BRANCH}" = "main" ]; then
+          EXTRA="--all"
+        fi
+        python3 .buildkite/common/scripts/upload_pipeline.py --upload $$EXTRA .buildkite/cuda/test-weekly.yml
     agents:
       queue: "cpu_queue_premerge"

.buildkite/cuda/scripts/check_vllm_runtime.py

@@ -0,0 +1,38 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+"""Verify the pinned vLLM and compiled CUDA imports on a GPU CI worker."""
+
+from __future__ import annotations
+
+import importlib
+import re
+from pathlib import Path
+
+
+def main() -> None:
+    dockerfile = Path(__file__).resolve().parents[3] / "docker" / "Dockerfile.ci"
+    match = re.search(
+        r"^ARG VLLM_VERSION=([0-9]+\.[0-9]+\.[0-9]+)$",
+        dockerfile.read_text(),
+        re.MULTILINE,
+    )
+    if match is None:
+        raise RuntimeError(f"Missing vLLM release version in {dockerfile}")
+    expected_version = match.group(1)
+
+    import torch
+    import vllm
+
+    assert vllm.__version__ == expected_version, vllm.__version__
+    assert torch.cuda.is_available(), "vLLM runtime check requires a GPU worker"
+    for module in ("flashinfer", "vllm._C_stable_libtorch"):
+        importlib.import_module(module)
+    print(
+        f"Verified vLLM CUDA runtime: vllm={vllm.__version__} "
+        f"expected={expected_version} torch={torch.__version__} "
+        f"cuda={torch.version.cuda} device={torch.cuda.get_device_name()}"
+    )
+
+
+if __name__ == "__main__":
+    main()

.buildkite/cuda/test-merge.yml

@@ -1,5 +1,6 @@
 # CUDA CI L3 (`advanced_model` e2e) plus L1 CPU Simple Test (`core_model and cpu`).
-# Uploaded by Upload Merge Pipeline: PR label `merge-test`, or post-merge on main.
+# Uploaded by Upload Merge Pipeline: PR label `merge-test`, or post-merge on main
+# (both diff-aware via source_file_dependencies).
 
 env:
   VLLM_WORKER_MULTIPROC_METHOD: spawn
@@ -17,7 +18,7 @@ steps:
         mirror_hardwares: l4_1
 
       - label: "Simple · Diffusion Test"
-        timeout_in_minutes: 20
+        timeout_in_minutes: 30
         commands:
           - "pytest -sv tests/diffusion -m 'core_model and cpu'"
         mirror_hardwares: l4_1
@@ -37,6 +38,17 @@ steps:
   - group: ":card_index_dividers: Diffusion Test"
     depends_on: upload-merge-pipeline
     steps:
+      - label: "Diffusion · AR Paged Attention GPU Test"
+        timeout_in_minutes: 20
+        commands:
+          - export VLLM_OMNI_AR_FA_REQUIRED=1
+          - >-
+            timeout 15m pytest -sv
+            tests/diffusion/ar_diffusion/test_paged_attention.py
+            -m 'core_model and cuda and L4 and cards_1'
+            --run-level "core_model"
+        mirror_hardwares: l4_1
+
       - label: "Diffusion · LoRA Test"
         timeout_in_minutes: 30
         commands:
@@ -59,40 +71,39 @@ steps:
     depends_on: upload-merge-pipeline
     steps:
       - label: "Entrypoint Test with H100"
-        # TEMPORARY 60 -> 90: engine startup/sleep-wake is ~13-15 min slower per
-        # run on the vLLM 0.26 image (model loading ~2x, init ~2x; see builds
-        # 2915 vs 2916), which pushed this suite past 60 min. Revert once the
-        # 0.26 startup regression is understood and fixed.
-        timeout_in_minutes: 90
+        timeout_in_minutes: 60
         commands:
-          - pytest -s -v tests/entrypoints/ -m "advanced_model and cuda and H100" --run-level "advanced_model"
+          # Turn-based server VAD over /v1/realtime is deferred to RFC #7181 PR 5
+          # (see the P0 thread on #7413); the test stays in the tree and runs again there.
+          - pytest -s -v tests/entrypoints/ -m "advanced_model and cuda and H100" --run-level "advanced_model" --deselect tests/entrypoints/openai_api/test_qwen3_omni_realtime_websocket.py::TestQwen3OmniRealtimeWebSocket::test_server_vad_multi_turn_without_client_commit
         mirror_hardwares: h100_2
 
       - label: "Entrypoint Test with L4"
         timeout_in_minutes: 60
+        # Workaround https://github.com/vllm-project/vllm-omni/issues/8016: shutdown()
+        # leaves inline UniProc DiffusionWorker VRAM mapped in the pytest parent.
+        # This job does not collect the ready leftover
+        # (test_diffusion_generate_request_id, core_model); isolate the
+        # advanced_model DiT handshake instead. Revert to a single sweep after
+        # the product shutdown fix.
         commands:
-          - pytest -s -v tests/entrypoints/ -m "advanced_model and cuda and L4" --run-level "advanced_model"
+          - |
+            set -eu
+            pytest -s -v tests/entrypoints/test_omni_sleep_mode.py::TestOmniDiffusionSleepMode::test_diffusion_sleep_handshake -m "advanced_model and cuda and L4 and cards_1" --run-level "advanced_model"
+            pytest -s -v tests/entrypoints/ -m "advanced_model and cuda and L4 and cards_1" --run-level "advanced_model" --deselect tests/entrypoints/test_omni_sleep_mode.py::TestOmniDiffusionSleepMode::test_diffusion_sleep_handshake
         mirror_hardwares: l4_1
 
+      - label: "Entrypoint Test with L4 · 2-GPU"
+        timeout_in_minutes: 60
+        commands:
+          - pytest -s -v tests/entrypoints/ -m "advanced_model and cuda and L4 and cards_2" --run-level "advanced_model"
+        mirror_hardwares: l4_2
+
   - group: ":card_index_dividers: E2E Test"
     depends_on: upload-merge-pipeline
     steps:
       - label: "TTS · Qwen3-TTS CustomVoice Test"
-        source_file_dependencies:
-          - tests/e2e/features/prefill_phase/
-          - vllm_omni/worker/gpu_model_runner.py
-          - tests/helpers/media.py
-          - tests/e2e/online_serving/test_qwen3_tts_customvoice.py
-          - tests/e2e/offline_inference/test_qwen3_tts_customvoice.py
-          - vllm_omni/model_executor/models/qwen3_tts/
-          - vllm_omni/model_executor/models/common/qwen3_code_predictor.py
-          - vllm_omni/model_executor/models/common/snake_activation.py
-          - vllm_omni/model_executor/models/common/alias_free_activation.py
-          - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
-          - vllm_omni/deploy/qwen3_tts.yaml
-          - vllm_omni/entrypoints/openai/serving_speech.py
-          - vllm_omni/entrypoints/openai/tts_adapters/
-          - vllm_omni/engine/stage_pool.py
+        source_file_dependencies: tts_qwen3_tts_function
         commands:
           - |
             timeout 40m bash -c "
@@ -103,21 +114,7 @@ steps:
         mirror_hardwares: l4_1
 
       - label: "TTS · Qwen3-TTS Base Test"
-        source_file_dependencies:
-          - tests/helpers/media.py
-          - tests/e2e/online_serving/test_qwen3_tts_base.py
-          - tests/e2e/offline_inference/test_qwen3_tts_base.py
-          - vllm_omni/model_executor/models/qwen3_tts/
-          - vllm_omni/model_executor/models/common/qwen3_code_predictor.py
-          - vllm_omni/model_executor/models/common/snake_activation.py
-          - vllm_omni/model_executor/models/common/alias_free_activation.py
-          - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
-          - vllm_omni/deploy/qwen3_tts.yaml
-          - vllm_omni/entrypoints/openai/serving_speech.py
-          - vllm_omni/entrypoints/openai/tts_adapters/
-          - vllm_omni/engine/stage_pool.py
-          - .buildkite/common/scripts/run_cov_split.sh
-          - pyproject.toml
+        source_file_dependencies: tts_qwen3_tts_cov
         # 40 -> 60: the coverage split runs the model twice.
         timeout_in_minutes: 60
         commands:
@@ -132,11 +129,23 @@ steps:
         mirror_hardwares: l4_1
 
       - label: "Omni · Qwen3-Omni Test"
+        source_file_dependencies: omni_qwen3_omni_function
+        timeout_in_minutes: 50
+        commands:
+          - pytest -s -v tests/e2e/offline_inference/test_qwen3_omni.py tests/e2e/online_serving/test_qwen3_omni.py tests/e2e/offline_inference/test_qwen3_omni_colocate_async.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+        mirror_hardwares: h100_2
+
+      - label: "Omni · Qwen3-Omni Async-Chunk SHM Leak"
         source_file_dependencies:
-          - tests/e2e/offline_inference/test_qwen3_omni.py
-          - tests/e2e/offline_inference/test_qwen3_omni_colocate_async.py
-          - tests/e2e/online_serving/test_qwen3_omni.py
+          - tests/e2e/online_serving/test_qwen3_omni_shm_leak.py
           - tests/helpers/runtime.py
+          - tests/helpers/stage_config.py
+          - vllm_omni/core/sched/omni_ar_scheduler.py
+          - vllm_omni/distributed/omni_connectors/connectors/shm_connector.py
+          - vllm_omni/distributed/omni_connectors/transfer_adapter/chunk_transfer_adapter.py
+          - vllm_omni/engine/orchestrator.py
+          - vllm_omni/engine/stage_engine_core_proc.py
+          - vllm_omni/engine/stage_pool.py
           - vllm_omni/entrypoints/openai/serving_chat.py
           - vllm_omni/model_executor/models/qwen3_omni/
           - vllm_omni/model_executor/models/common/qwen3_code_predictor.py
@@ -145,23 +154,13 @@ steps:
           - vllm_omni/model_executor/models/qwen3_tts/cuda_graph_decoder_wrapper.py
           - vllm_omni/model_executor/stage_input_processors/qwen3_omni.py
           - vllm_omni/deploy/qwen3_omni_moe.yaml
-          - vllm_omni/deploy/qwen3_omni_moe_thinking.yaml
-        timeout_in_minutes: 50
+        timeout_in_minutes: 40
         commands:
-          - pytest -s -v tests/e2e/offline_inference/test_qwen3_omni.py tests/e2e/online_serving/test_qwen3_omni.py tests/e2e/offline_inference/test_qwen3_omni_colocate_async.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+          - pytest -s -v tests/e2e/online_serving/test_qwen3_omni_shm_leak.py -m 'advanced_model and cuda' --run-level 'advanced_model'
         mirror_hardwares: h100_2
 
       - label: "Omni · MiniCPM-o 4.5 Test"
-        source_file_dependencies:
-          - tests/e2e/offline_inference/test_minicpmo_4_5.py
-          - tests/e2e/online_serving/test_minicpmo_4_5.py
-          - vllm_omni/transformers_utils/chat_templates/minicpmo45_native.jinja
-          - vllm_omni/model_executor/models/minicpmo_4_5/
-          - vllm_omni/model_executor/models/step_audio2/
-          - vllm_omni/model_executor/models/cosyvoice3/code2wav_core/
-          - vllm_omni/model_executor/models/cosyvoice3/utils.py
-          - vllm_omni/model_executor/stage_input_processors/minicpmo_4_5_omni.py
-          - vllm_omni/deploy/minicpmo_4_5.yaml
+        source_file_dependencies: omni_minicpmo_4_5_function
         timeout_in_minutes: 50
         commands:
           - pytest -s -v tests/e2e/offline_inference/test_minicpmo_4_5.py tests/e2e/online_serving/test_minicpmo_4_5.py -m 'advanced_model and cuda' --run-level 'advanced_model'
@@ -174,59 +173,14 @@ steps:
         mirror_hardwares: h100_2
 
       - label: "Omni · MiniCPM-o 4.5 Duplex Test"
-        source_file_dependencies:
-          - tests/assets/minicpmo_4_5/response_required_16k.wav
-          - tests/assets/minicpmo_4_5/soft_interrupt_16k.wav
-          - tests/e2e/online_serving/test_minicpmo_4_5_duplex.py
-          - tests/e2e/online_serving/test_minicpmo_4_5_duplex_expansion.py
-          - tests/e2e/online_serving/test_minicpmo_realtime_duplex_drivers.py
-          - tests/e2e/online_serving/helpers/minicpmo_realtime_duplex_scenarios.py
-          - tests/e2e/online_serving/run_minicpmo_realtime_duplex_multi_session.py
-          - tests/e2e/online_serving/run_minicpmo_realtime_duplex_server_vad.py
-          - tests/e2e/online_serving/run_minicpmo_realtime_duplex_soft_interrupt.py
-          - vllm_omni/deploy/minicpmo_4_5.yaml
-          - tests/e2e/online_serving/helpers/minicpmo_4_5_duplex.py
-          - vllm_omni/engine/duplex/
-          - vllm_omni/entrypoints/duplex/
-          # Public duplex client library + its unit suites: client-side
-          # changes must re-run the duplex e2e group. The live client probe
-          # (test_duplex_client_live.py, single GPU, core tier) runs in the
-          # ready pipeline's duplex group.
-          - vllm_omni/clients/
-          - tests/clients/
-          - vllm_omni/entrypoints/duplex_request_client.py
-          - vllm_omni/outputs/duplex.py
-          - vllm_omni/model_executor/duplex_sampling.py
-          - tests/engine/duplex/
-          - tests/entrypoints/duplex/
-          - tests/model_executor/models/minicpmo_4_5/duplex/
-          # Legacy Realtime probe client + video stacking used by the e2e drivers.
-          - vllm_omni/experimental/fullduplex/
-          - vllm_omni/model_executor/models/minicpmo_4_5/
-          - vllm_omni/model_executor/models/step_audio2/
-          - vllm_omni/model_executor/models/cosyvoice3/code2wav_core/
-          - vllm_omni/model_executor/models/cosyvoice3/utils.py
-          - vllm_omni/model_executor/stage_input_processors/minicpmo_4_5_omni.py
-          - vllm_omni/core/sched/
-          - vllm_omni/outputs/multimodal_accumulation.py
-          - vllm_omni/engine/stage_init_utils.py
-          - vllm_omni/engine/async_omni_engine.py
-          - vllm_omni/engine/orchestrator.py
-          - vllm_omni/entrypoints/cli/serve.py
-          - vllm_omni/entrypoints/openai/api_server.py
-          - vllm_omni/entrypoints/openai/duplex_capability.py
-          - vllm_omni/config/stage_config.py
+        source_file_dependencies: omni_minicpmo_4_5_duplex_function
         timeout_in_minutes: 50
         commands:
-          - pytest -s -v tests/e2e/online_serving/test_minicpmo_4_5_duplex.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+          - pytest -s -v tests/e2e/online_serving/test_minicpmo_4_5_duplex.py tests/e2e/online_serving/test_minicpmo_4_5_window.py -m 'advanced_model and cuda' --run-level 'advanced_model'
         mirror_hardwares: h100_1
 
       - label: "Diffusion · Qwen Image Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_qwen_image.py
-          - vllm_omni/diffusion/models/qwen_image/
-          - vllm_omni/diffusion/models/dmd2/
-          - vllm_omni/diffusion/distributed/autoencoders/autoencoder_kl_qwenimage.py
+        source_file_dependencies: diffusion_qwen_image_function
         commands:
           - |
             timeout 40m bash -c '
@@ -235,12 +189,17 @@ steps:
             '
         mirror_hardwares: h100_1
 
+      - label: "Diffusion · Kandinsky6 Test"
+        source_file_dependencies: diffusion_kandinsky6_function
+        commands:
+          - |
+            timeout 40m bash -c "
+              pytest -s -v tests/e2e/offline_inference/test_kandinsky6.py tests/e2e/online_serving/test_kandinsky6.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
+            "
+        mirror_hardwares: h100_1
+
       - label: "Diffusion · Wan22 Test"
-        source_file_dependencies:
-          - tests/e2e/offline_inference/test_wan22_t2v.py
-          - tests/e2e/online_serving/test_wan22_t2v.py
-          - vllm_omni/diffusion/models/wan2_2/
-          - vllm_omni/diffusion/models/dmd2/
+        source_file_dependencies: diffusion_wan22_function
         commands:
           - |
             timeout 40m bash -c "
@@ -250,9 +209,7 @@ steps:
         mirror_hardwares: h100_1
 
       - label: "Diffusion · HunyuanVideo-1.5 Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_hunyuan_video_15_expansion.py
-          - vllm_omni/diffusion/models/hunyuan_video/
+        source_file_dependencies: diffusion_hunyuan_video_function
         commands:
           - |
             timeout 40m bash -c "
@@ -261,18 +218,18 @@ steps:
             "
         mirror_hardwares: h100_1
 
+      - label: "Diffusion · LingBot-Video Test"
+        source_file_dependencies: diffusion_lingbot_function
+        commands:
+          - |
+            timeout 40m bash -c "
+              export VLLM_IMAGE_FETCH_TIMEOUT=60
+              pytest -s -v tests/e2e/online_serving/test_lingbot_video.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
+            "
+        mirror_hardwares: h100_1
+
       - label: "Diffusion · MiniMax-H3 L3 DLO DP2 and FastH3"
-        source_file_dependencies:
-          - tests/e2e/online_serving/minimax_h3/
-          - tests/helpers/assertions.py
-          - tests/helpers/mark.py
-          - tests/helpers/media.py
-          - tests/helpers/runtime.py
-          - vllm_omni/diffusion/models/minimax_h3/
-          - vllm_omni/diffusion/lora/
-          - vllm_omni/diffusion/offloader/
-          - vllm_omni/diffusion/distributed/
-          - vllm_omni/entrypoints/openai/
+        source_file_dependencies: diffusion_minimax_h3_function
         # Keep four server lifecycles in one Buildkite job, but never in one
         # pytest process. DLO uses multiprocessing shared-memory queues; the
         # parent's resource_tracker can retain a stale /psm_* entry during
@@ -295,3 +252,14 @@ steps:
             timeout 35m pytest -s -v tests/e2e/online_serving/minimax_h3/test_minimax_h3_turbo_lora.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
             timeout 35m pytest -s -v tests/e2e/online_serving/minimax_h3/test_minimax_h3_fasth3_lora.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
         mirror_hardwares: h100_2
+
+      - label: "Diffusion · MiniMax-H3 Latent Upscale and Refine"
+        source_file_dependencies: diffusion_minimax_h3_function
+        timeout_in_minutes: 40
+        commands:
+          - |
+            set -eu
+            VLLM_TEST_MINIMAX_H3_LATENT_UPSCALER=$(python3 -c 'from huggingface_hub import hf_hub_download; print(hf_hub_download(repo_id="LBH-123-AI/Minimax_h3_latent_Upscaler", filename="minimax_h3_latent_upscaler_3d_conv_v1/minimax_h3_latent_upscaler_3d_conv_v1_bf16.safetensors"))')
+            export VLLM_TEST_MINIMAX_H3_LATENT_UPSCALER
+            timeout 35m pytest -s -v tests/e2e/online_serving/minimax_h3/test_minimax_h3.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
+        mirror_hardwares: h100_2

.buildkite/cuda/test-nightly.yml

@@ -12,164 +12,184 @@ steps:
   - group: ":card_index_dividers: Omni Model Test"
     key: nightly-omni-test-group
     depends_on: upload-nightly-pipeline
-    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test" || build.pull_request.labels includes "omni-test"
+    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test"
     steps:
 
       - label: ":full_moon: Omni · Function Test with H100 · Single-GPU"
+        source_file_dependencies:
+          - omni_qwen3_omni_function
+          - omni_minicpmo_4_5_function
         timeout_in_minutes: 120
         commands:
-          - pytest -sv tests/e2e --run-level "full_model" -m "full_model and H100 and B200 and omni and cards_1" --ignore=tests/e2e/accuracy --ignore=tests/e2e/online_serving/test_minicpmo_4_5_duplex_expansion.py
+          - pytest -sv tests/e2e --run-level "full_model" -m "full_model and H100 and B200 and omni and cards_1" --ignore=tests/e2e/accuracy --ignore=tests/e2e/online_serving/test_minicpmo_4_5_duplex_expansion.py --ignore=tests/e2e/online_serving/test_minicpmo_4_5_window.py
 
       - label: ":full_moon: Omni · Function Test with H100 · 2-GPU"
+        source_file_dependencies:
+          - omni_qwen3_omni_function
+          - omni_minicpmo_4_5_function
         timeout_in_minutes: 90
         commands:
           - pytest -sv tests/e2e --run-level "full_model" -m "full_model and H100 and B200 and omni and cards_2" --ignore=tests/e2e/accuracy --ignore=tests/e2e/online_serving/test_qwen3_omni_multi_replicas.py
 
       - label: ":full_moon: Omni · MiniCPM-o 4.5 Duplex Test"
-        timeout_in_minutes: 50
+        source_file_dependencies: omni_minicpmo_4_5_duplex_function
+        timeout_in_minutes: 120
         commands:
-          - pytest -s -v tests/e2e/online_serving/test_minicpmo_4_5_duplex_expansion.py -m "full_model and cuda and H100 and omni and cards_1" --run-level "full_model"
-        mirror_hardwares: h100_1
+          - pytest -s -v tests/e2e/online_serving/test_minicpmo_4_5_duplex_expansion.py -m "full_model and cuda and H100 and B200 and omni and cards_1" --run-level "full_model"
+          - pytest -s -v tests/e2e/online_serving/test_minicpmo_4_5_window.py -m "full_model and cuda and H100 and B200 and omni and cards_1" --run-level "full_model"
 
       - label: ":full_moon: Omni · Doc Test with L4 · 4-GPU"
+        source_file_dependencies: omni_qwen2_5_omni_doc
         timeout_in_minutes: 90
         commands:
           - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-          - pytest -s -v tests/examples/ -m "full_model and omni and L4 and cards_4" --run-level "full_model"
-        mirror_hardwares: l4_4
+          - pytest -s -v tests/examples/ -m "full_model and omni and L4 and B200 and cards_4" --run-level "full_model"
 
       - label: ":full_moon: Omni · Doc Test with H100 · 2-GPU"
+        source_file_dependencies: omni_qwen3_omni_doc
         timeout_in_minutes: 90
+        artifact_paths:
+          - audio_*.wav
         commands:
-          - pytest -s -v tests/examples/ -m "full_model and omni and H100 and cards_2" --run-level "full_model"
-        mirror_hardwares: h100_2
+          - pytest -s -v tests/examples/ -m "full_model and omni and H100 and B200 and cards_2" --run-level "full_model"
 
       - label: ":full_moon: Omni · Accuracy Test"
+        source_file_dependencies: omni_qwen3_omni_accuracy
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/e2e/accuracy/qwen3_omni/results/qwen_omni_acc/*.json
         commands:
           - export SEED_TTS_WER_EVAL=1
           - export SEED_TTS_EVAL_DEVICE=cuda:1
-          - |
-            set +e
-            pytest -s -v tests/e2e/accuracy/qwen3_omni/test_qwen3_omni.py -m "full_model" --run-level full_model
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/e2e/accuracy/qwen3_omni/results/qwen_omni_acc/*.json"
-            exit $$EXIT
-        mirror_hardwares: h100_2
+          - pytest -s -v tests/e2e/accuracy/qwen3_omni/test_qwen3_omni.py -m "full_model and H100 and B200 and cards_2" --run-level full_model
 
       - label: ":full_moon: Omni · MiniCPM-o 4.5 · Accuracy Test"
+        source_file_dependencies: omni_minicpmo_4_5_accuracy
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/e2e/accuracy/minicpmo_4_5/results/*.json
         commands:
           - export SEED_TTS_WER_EVAL=1
           - export SEED_TTS_EVAL_DEVICE=cuda:1
+          - pytest -s -v tests/e2e/accuracy/minicpmo_4_5/test_minicpmo_4_5.py -m "full_model and H100 and B200 and cuda and (cards_1 or cards_2)" --run-level full_model
+
+      - label: ":full_moon: Omni · Omni-DuplexEval CI Guard"
+        source_file_dependencies: omni_duplex_eval_ci
+        timeout_in_minutes: 240
+        env:
+          VLLM_DUPLEX_EVAL_JUDGE_URL: "http://127.0.0.1:8001"
+        commands:
           - |
             set +e
-            pytest -s -v tests/e2e/accuracy/minicpmo_4_5/test_minicpmo_4_5.py -m "full_model and cuda" --run-level full_model
+            pytest -s -v tests/e2e/accuracy/omni_duplex_eval/test_omni_duplex_eval_ci.py \
+              -m "full_model and H100 and B200 and cuda and cards_2" \
+              --run-level full_model \
+              --basetemp /tmp/omni_duplex_ci
             EXIT=$$?
-            buildkite-agent artifact upload "tests/e2e/accuracy/minicpmo_4_5/results/*.json"
+            buildkite-agent artifact upload "tests/e2e/accuracy/omni_duplex_eval/results/*.json"
             exit $$EXIT
-        mirror_hardwares: h100_2
 
       - label: ":full_moon: Omni · Perf Test · No Async Chunk"
+        source_file_dependencies: omni_qwen3_omni_perf
         key: nightly-omni-performance-no-async-chunk
         timeout_in_minutes: 300
+        artifact_paths:
+          - tests/dfx/perf/results/*.json
         commands:
           - export BENCHMARK_DIR=tests/dfx/perf/results
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen3_omni_no_async_chunk.json
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
-            exit $$EXIT
-        mirror_hardwares: h100_2
+          - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen3_omni_no_async_chunk.json -m "H100 and B200 and cards_2"
 
       - label: ":full_moon: Omni · Perf Test · Async Chunk"
+        source_file_dependencies: omni_qwen3_omni_perf
         key: nightly-omni-performance-async-chunk
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/dfx/perf/results/*.json
         commands:
           - export BENCHMARK_DIR=tests/dfx/perf/results
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen3_omni_async_chunk.json -m "H100 and full_model"
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
-            exit $$EXIT
-        mirror_hardwares: h100_2
+          - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen3_omni_async_chunk.json -m "H100 and B200 and full_model and cards_2"
 
       - label: ":full_moon: Omni · MiniCPM-o 4.5 · Perf Test"
+        source_file_dependencies: omni_minicpmo_4_5_perf
         key: nightly-omni-performance-minicpmo-4-5
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/dfx/perf/results/*.json
         commands:
           - export BENCHMARK_DIR=tests/dfx/perf/results
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_minicpmo_4_5.json
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
-            exit $$EXIT
-        mirror_hardwares: h100_1
+          - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_minicpmo_4_5.json -m "H100 and B200 and cards_1"
 
       - label: ":full_moon: Omni · MiniCPM-o 4.5 · Duplex Seed-TTS Perf Test"
+        source_file_dependencies: omni_minicpmo_4_5_duplex_perf
         key: nightly-omni-performance-minicpmo-4-5-duplex-seed-tts
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/dfx/perf/results/*.json
         commands:
           - export BENCHMARK_DIR=tests/dfx/perf/results
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_minicpmo_4_5_duplex_seed_tts.json
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
-            exit $$EXIT
-        mirror_hardwares: h100_1
+          - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_minicpmo_4_5_duplex_seed_tts.json -m "H100 and B200 and cards_1"
 
       - label: ":full_moon: Omni · Multi-Replica Startup Test with 4x H100"
+        source_file_dependencies: omni_qwen3_omni_function
         timeout_in_minutes: 45
         commands:
-          - pytest -s -v tests/e2e/online_serving/test_qwen3_omni_multi_replicas.py -m "full_model" --run-level "core_model"
-        mirror_hardwares: h100_4
+          - pytest -s -v tests/e2e/online_serving/test_qwen3_omni_multi_replicas.py -m "full_model and H100 and B200 and cards_4" --run-level "core_model"
+
+  - group: ":card_index_dividers: Entrypoints Test"
+    key: nightly-entrypoints-test-group
+    depends_on: upload-nightly-pipeline
+    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test"
+    steps:
+
+      - label: ":full_moon: Entrypoint Test with H100"
+        source_file_dependencies: entrypoints_function
+        # Merge collects advanced_model (5 duplex interruptions). Nightly
+        # collects full_model (20 interruptions + BAGEL sleep).
+        timeout_in_minutes: 180
+        commands:
+          - pytest -s -v tests/entrypoints/ -m "full_model and cuda and H100 and B200 and cards_2" --run-level "full_model"
 
   - group: ":card_index_dividers: TTS Model Test"
     key: nightly-tts-test-group
     depends_on: upload-nightly-pipeline
-    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test" || build.pull_request.labels includes "tts-test"
+    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test"
     steps:
 
       - label: ":full_moon: TTS · Function Test with L4"
+        source_file_dependencies: tts_qwen3_tts_function
         timeout_in_minutes: 120
+        env:
+          VLLM_OMNI_FAILED_SPEECH_AUDIO_DIR: .
+        artifact_paths:
+          - failed_speech_*.wav
+          - failed_speech_*.json
         commands:
           - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
           - export VLLM_USE_DEEP_GEMM="0"
           - export VLLM_MOE_USE_DEEP_GEMM="0"
-          - pytest -s -v tests/e2e/ -m "full_model and L4 and tts" --run-level "full_model" --ignore=tests/e2e/accuracy
-        mirror_hardwares: l4_1
+          - pytest -s -v tests/e2e/ -m "full_model and L4 and B200 and tts and cards_1" --run-level "full_model" --ignore=tests/e2e/accuracy
 
       - label: ":full_moon: TTS · Perf Test · Single-GPU"
+        source_file_dependencies: tts_qwen3_tts_perf
         key: nightly-tts-performance-single
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/dfx/perf/results/*.json
         commands:
           - export BENCHMARK_DIR=tests/dfx/perf/results
           - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_tts.json -m "H100 and cards_1"
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
-            exit $$EXIT
-        mirror_hardwares: h100_1
+          - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_tts.json -m "H100 and B200 and cards_1"
 
       - label: ":full_moon: TTS · Perf Test · 2-GPU"
+        source_file_dependencies: tts_qwen3_tts_perf
         key: nightly-tts-performance-2gpu
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/dfx/perf/results/*.json
         commands:
           - export BENCHMARK_DIR=tests/dfx/perf/results
           - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_tts.json -m "H100 and cards_2"
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
-            exit $$EXIT
-        mirror_hardwares: h100_2
+          - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_tts.json -m "H100 and B200 and cards_2"
 
   - group: ":card_index_dividers: Tiny Model Tests [Multi-GPU]"
     key: nightly-tiny-model-test-group
@@ -180,6 +200,7 @@ steps:
     steps:
 
       - label: ":full_moon: Tiny Model Tests · Diffusion (Multi-GPU accelerations) · 2-GPU"
+        source_file_dependencies: diffusion_tiny_model
         timeout_in_minutes: 30
         # NOTE: Single-GPU (core_model) tiny model tests run on the ready label,
         # so nightly only adds the multi-GPU parallelism configurations.
@@ -189,260 +210,259 @@ steps:
         # for the CI, since these tests should pass irrespective of whether or not we use
         # the full weights or not.
         commands:
-          - pytest -sv tests/model_tests/diffusion -m 'full_model and cuda and cards_2' --run-level "core_model"
-        mirror_hardwares: l4_2
+          - pytest -sv tests/model_tests/diffusion -m 'full_model and L4 and B200 and cuda and cards_2' --run-level "core_model"
 
       - label: ":full_moon: Tiny Model Tests · Diffusion (Multi-GPU accelerations) · 4-GPU"
+        source_file_dependencies: diffusion_tiny_model
         timeout_in_minutes: 30
         # CFG+TP extra_test_groups request 2×2=4 devices (see get_required_device_count).
         commands:
-          - pytest -sv tests/model_tests/diffusion -m 'full_model and cuda and cards_4' --run-level "core_model"
-        mirror_hardwares: l4_4
+          - pytest -sv tests/model_tests/diffusion -m 'full_model and L4 and B200 and cuda and cards_4' --run-level "core_model"
 
   - group: ":card_index_dividers: Diffusion X2I(&A&T) Model Test"
     key: nightly-diffusion-x2iat-group
     depends_on: upload-nightly-pipeline
     if: >-
       build.env("NIGHTLY") == "1" ||
-      build.pull_request.labels includes "nightly-test" ||
-      build.pull_request.labels includes "diffusion-x2iat-test"
+      build.pull_request.labels includes "nightly-test"
     steps:
 
       - label: ":full_moon: Diffusion X2I(&A&T) · Function Test with H100 · Single-GPU"
+        source_file_dependencies: diffusion_qwen_image_function
         timeout_in_minutes: 120
         commands:
           - >-
             pytest -sv
             tests/e2e/online_serving/test_qwen_image_expansion.py
-            -m "full_model and diffusion and H100 and cards_1"
+            -m "full_model and diffusion and H100 and B200 and cards_1"
             --run-level "full_model"
-        mirror_hardwares: h100_1
 
       - label: ":full_moon: Diffusion X2I(&A&T) · Function Test with H100 · 2-GPU"
+        source_file_dependencies: diffusion_qwen_image_function
         timeout_in_minutes: 120
         commands:
           - >-
             pytest -sv
             tests/e2e/online_serving/test_qwen_image_expansion.py
-            -m "full_model and diffusion and H100 and cards_2"
+            -m "full_model and diffusion and H100 and B200 and cards_2"
             --run-level "full_model"
-        mirror_hardwares: h100_2
 
-      - label: ":full_moon: Diffusion X2I(&A&T) · Boogu Edit CFG · 3-GPU"
+      - label: ":full_moon: Diffusion X2I(&A&T) · Function Test with L4 · Single-GPU"
+        source_file_dependencies: diffusion_qwen_image_function
         timeout_in_minutes: 120
         commands:
-          - >-
-            pytest -sv
-            tests/e2e/online_serving/test_boogu_image_edit_expansion.py
-            -k "test_double_guidance_cfg_parallel and cfg_parallel_3_full"
-            -m "slow and diffusion and cuda and H100 and cards_3"
-            --run-level "full_model"
-        mirror_hardwares: h100_3
+          - pytest -sv tests/e2e/offline_inference/test_qwen_image_autoround_w4a16_expansion.py -m "full_model and diffusion and L4 and B200 and cards_1" --run-level "full_model"
 
-      - label: ":full_moon: Diffusion X2I(&A&T) · Function Test with L4 · 4-GPU"
-        timeout_in_minutes: 120
-        commands:
-          - pytest -sv tests/e2e/ -k "not test_wan and not hunyuan" -m "full_model and diffusion and L4 and cards_4" --run-level "full_model" --ignore=tests/e2e/accuracy
-        mirror_hardwares: l4_4
 
       - label: ":full_moon: Diffusion X2I(&A&T) · Doc Test"
+        source_file_dependencies: diffusion_text_to_image_doc
         timeout_in_minutes: 60
         commands:
-          - pytest -s -v tests/examples/*/test_text_to_image.py -m "full_model and example and H100" --run-level "full_model"
-        mirror_hardwares: h100_2
+          - pytest -s -v tests/examples/*/test_text_to_image.py -m "full_model and example and H100 and B200 and (cards_1 or cards_2)" --run-level "full_model"
 
       - label: ":full_moon: Diffusion X2I(&A&T) · Accuracy Test"
+        source_file_dependencies: diffusion_qwen_image_accuracy
         timeout_in_minutes: 180
         commands:
           - export VLLM_HTTP_TIMEOUT_KEEP_ALIVE=120
-          - pytest -s -v tests/e2e/accuracy/test_qwen_image.py --run-level full_model
-          - pytest -s -v tests/e2e/accuracy/test_diffusers_backend_similarity.py -k '2i_matches_diffusers' -m full_model --run-level full_model
-        mirror_hardwares: h100_1
+          - pytest -s -v tests/e2e/accuracy/test_qwen_image.py -m "H100 and B200 and cards_1" --run-level full_model
+          - pytest -s -v tests/e2e/accuracy/test_diffusers_backend_similarity.py -k '2i_matches_diffusers' -m "full_model and H100 and B200 and cards_1" --run-level full_model
+
+      - label: ":full_moon: JoyAI-Image-Edit · Accuracy Test"
+        source_file_dependencies: diffusion_joy_image_accuracy
+        timeout_in_minutes: 180
+        commands:
+          - pytest -s -v tests/e2e/accuracy/test_joy_image_edit.py -m "full_model and diffusion and H100 and B200 and cards_1" --run-level full_model
 
       - label: ":full_moon: HunyuanImage3-DIT · Accuracy Test"
+        source_file_dependencies: diffusion_hunyuan_image3_accuracy
         timeout_in_minutes: 180
         commands:
           - export HUNYUAN_IMAGE3_MODEL="tencent/HunyuanImage-3.0-Instruct"
           - export HUNYUAN_IMAGE3_DEVICES="0,1,2,3"
-          - pytest -s -v tests/e2e/accuracy/test_hunyuan_image3_pixel_accuracy.py -m full_model --run-level full_model
-        mirror_hardwares: h100_4
-
-      - label: ":full_moon: HiDream-O1-Image · Accuracy Test"
-        timeout_in_minutes: 180
-        commands:
-          - pytest -s -v tests/e2e/accuracy/test_hidream_o1_image.py -m full_model --run-level full_model
-        mirror_hardwares: h100_2
+          - pytest -s -v tests/e2e/accuracy/test_hunyuan_image3_pixel_accuracy.py tests/e2e/accuracy/test_hunyuan_image3_lora.py -m "full_model and H100 and B200 and cards_4" --run-level full_model
 
       - label: ":full_moon: Diffusion X2I(&A&T) · Perf Test · Qwen-Image · Single-GPU"
+        source_file_dependencies: diffusion_qwen_image_perf
         key: nightly-diffusion-x2iat-performance-qwen-image-single
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/dfx/perf/results/diffusion_result_*.json
+          - tests/dfx/perf/results/logs/*.log
         commands:
           - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
           - export CACHE_DIT_VERSION=1.5.0
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen_image_vllm_omni.json -m "H100 and cards_1"
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
-            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
-            exit $$EXIT
-        mirror_hardwares: h100_1
+          - pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen_image_vllm_omni.json -m "H100 and B200 and cards_1"
 
       - label: ":full_moon: Diffusion X2I(&A&T) · Perf Test · Qwen-Image · 4-GPU"
+        source_file_dependencies: diffusion_qwen_image_perf
         key: nightly-diffusion-x2iat-performance-qwen-image-4gpu
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/dfx/perf/results/diffusion_result_*.json
+          - tests/dfx/perf/results/logs/*.log
         commands:
           - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
-          - export DIFFUSION_ATTENTION_BACKEND=FLASH_ATTN
           - export CACHE_DIT_VERSION=1.5.0
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen_image_vllm_omni.json -m "H100 and cards_4"
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
-            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
-            exit $$EXIT
-        mirror_hardwares: h100_4
+          # Do not pin DIFFUSION_ATTENTION_BACKEND: H100 auto-selects FLASH_ATTN;
+          # B200 rejects explicit FLASH_ATTN without FA4 and uses CUDNN/TRTLLM.
+          - pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen_image_vllm_omni.json -m "H100 and B200 and cards_4"
 
   - group: ":card_index_dividers: Diffusion X2V Model Test"
     key: nightly-diffusion-x2v-group
     depends_on: upload-nightly-pipeline
     if: >-
       build.env("NIGHTLY") == "1" ||
-      build.pull_request.labels includes "nightly-test" ||
-      build.pull_request.labels includes "diffusion-x2v-test"
+      build.pull_request.labels includes "nightly-test"
     steps:
 
       - label: ":full_moon: Diffusion X2V · Wan2.2 T2V Function Test · Single-GPU"
+        source_file_dependencies: diffusion_wan22_function
         timeout_in_minutes: 90
         commands:
-          - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py -k "t2v" -m "full_model and cuda and H100 and cards_1" --run-level "full_model"
-        mirror_hardwares: h100_1
+          - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py -k "t2v" -m "full_model and cuda and H100 and B200 and cards_1" --run-level "full_model"
 
       - label: ":full_moon: Diffusion X2V · Wan2.2 T2V Function Test · 2-GPU"
+        source_file_dependencies: diffusion_wan22_function
         timeout_in_minutes: 90
         commands:
-          - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py -k "t2v" -m "full_model and cuda and H100 and cards_2" --run-level "full_model"
-        mirror_hardwares: h100_2
+          - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py -k "t2v" -m "full_model and cuda and H100 and B200 and cards_2" --run-level "full_model"
 
       - label: ":full_moon: Diffusion X2V · Wan2.2 I2V/TI2V Function Test · Single-GPU"
+        source_file_dependencies: diffusion_wan22_function
         timeout_in_minutes: 90
         commands:
-          - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py -k "i2v" -m "full_model and cuda and H100 and cards_1" --run-level "full_model"
-        mirror_hardwares: h100_1
+          - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py -k "i2v" -m "full_model and cuda and H100 and B200 and cards_1" --run-level "full_model"
 
       - label: ":full_moon: Diffusion X2V · Wan2.2 I2V/TI2V Function Test · 2-GPU"
+        source_file_dependencies: diffusion_wan22_function
         timeout_in_minutes: 90
         commands:
-          - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py -k "i2v" -m "full_model and cuda and H100 and cards_2" --run-level "full_model"
-        mirror_hardwares: h100_2
+          - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py -k "i2v" -m "full_model and cuda and H100 and B200 and cards_2" --run-level "full_model"
 
       - label: ":full_moon: Diffusion X2V · HunyuanVideo-1.5 Function Test"
+        source_file_dependencies: diffusion_hunyuan_video_function
         timeout_in_minutes: 90
         commands:
-          - pytest -s -v tests/e2e/online_serving/test_hunyuan_video_15_expansion.py -m "full_model and cuda" --run-level "full_model"
-        mirror_hardwares: h100_2
+          - pytest -s -v tests/e2e/online_serving/test_hunyuan_video_15_expansion.py -m "full_model and H100 and B200 and cuda and (cards_1 or cards_2)" --run-level "full_model"
 
-      - label: ":full_moon: Diffusion X2V · MAGI-2 Preview Serving Smoke"
-        timeout_in_minutes: 180
+      - label: ":full_moon: Diffusion X2V · LingBot Function Test"
+        source_file_dependencies: diffusion_lingbot_function
+        timeout_in_minutes: 120
         commands:
-          - pytest -s -v tests/e2e/online_serving/test_magi2.py -m "advanced_model and diffusion and cuda and H100 and cards_4" --run-level "advanced_model"
-        mirror_hardwares: h100_4
+          - >-
+            pytest -s -v
+            tests/e2e/online_serving/test_lingbot_video_expansion.py
+            tests/e2e/online_serving/test_lingbot_video_moe.py
+            tests/e2e/offline_inference/test_lingbot_world_v2.py
+            -m "full_model and diffusion and H100 and B200 and cards_1"
+            --run-level "full_model"
 
       - label: ":full_moon: Diffusion X2V · Doc Test"
+        source_file_dependencies: diffusion_image_to_video_doc
         timeout_in_minutes: 60
         commands:
-          - pytest -s -v tests/examples/offline_inference/test_image_to_video.py -m "full_model and example and H100" --run-level "full_model"
-        mirror_hardwares: h100_1
+          - pytest -s -v tests/examples/offline_inference/test_image_to_video.py -m "full_model and example and H100 and B200 and cards_1" --run-level "full_model"
 
-      - label: ":full_moon: Diffusion X2V · Other Function Test"
+      - label: ":full_moon: Diffusion X2V · Wan2.2 Autoround Function Test"
+        source_file_dependencies: diffusion_wan22_function
         timeout_in_minutes: 90
         commands:
-          - pytest -s -v tests/e2e/offline_inference/test_wan22_autoround_w4a16_expansion.py -m "full_model and cuda" --run-level "full_model"
-        mirror_hardwares: h100_2
+          - pytest -s -v tests/e2e/offline_inference/test_wan22_autoround_w4a16_expansion.py -m "full_model and H100 and B200 and cuda and (cards_1 or cards_2)" --run-level "full_model"
+
+      - label: ":full_moon: Diffusion X2V · Wan2.2 Device-Postprocess Equivalence Test"
+        source_file_dependencies: diffusion_wan22_function
+        timeout_in_minutes: 90
+        commands:
+          - |
+            VLLM_OMNI_DEVICE_POSTPROCESS_MODEL=Wan-AI/Wan2.2-TI2V-5B-Diffusers \
+              pytest -s -v tests/e2e/offline_inference/test_device_postprocess_equivalence.py -m "full_model and H100 and B200 and cuda and cards_1" --run-level "full_model"
 
       - label: ":full_moon: Diffusion X2V · Wan2.2 Accuracy Test"
+        source_file_dependencies: diffusion_wan22_accuracy
         timeout_in_minutes: 180
         commands:
-          - pytest -s -v tests/e2e/accuracy/wan22_i2v/test_wan22_i2v_video_similarity.py -m full_model --run-level full_model
-          - pytest -s -v tests/e2e/accuracy/test_diffusers_backend_similarity.py -k '2v_matches_diffusers' -m full_model --run-level full_model
-        mirror_hardwares: h100_2
+          - pytest -s -v tests/e2e/accuracy/wan22_i2v/test_wan22_i2v_video_similarity.py -m "full_model and H100 and B200 and (cards_1 or cards_2)" --run-level full_model
+          - pytest -s -v tests/e2e/accuracy/test_diffusers_backend_similarity.py -k '2v_matches_diffusers' -m "full_model and H100 and B200 and cards_1" --run-level full_model
 
       - label: ":full_moon: Diffusion X2V · MiniMax H3 I2VA/Ref2VA Accuracy Test"
+        source_file_dependencies: diffusion_minimax_h3_accuracy
         timeout_in_minutes: 180
         commands:
-          - pytest -s -v tests/e2e/accuracy/minimax_h3/test_minimax_h3_i2va_ref2va_similarity.py -m full_model --run-level full_model
-        mirror_hardwares: h100_4
+          - pytest -s -v tests/e2e/accuracy/minimax_h3/test_minimax_h3_i2va_ref2va_similarity.py -m "full_model and H100 and B200 and cards_4" --run-level full_model
 
-      - label: ":full_moon: Diffusion X2V · Other Accuracy Test"
+      - label: ":full_moon: Diffusion X2V · HunyuanVideo-1.5 Accuracy Test"
+        source_file_dependencies: diffusion_hunyuan_video_accuracy
         timeout_in_minutes: 180
         commands:
-          - pytest -s -v tests/e2e/accuracy/hunyuanvideo15_t2v/test_hunyuanvideo15_t2v_video_similarity.py -m full_model --run-level full_model
-          - pytest -s -v tests/e2e/accuracy/hunyuanvideo15_i2v/test_hunyuanvideo15_i2v_video_similarity.py -m full_model --run-level full_model
-        mirror_hardwares: h100_2
+          - pytest -s -v tests/e2e/accuracy/hunyuanvideo15_t2v/test_hunyuanvideo15_t2v_video_similarity.py -m "full_model and H100 and B200 and (cards_1 or cards_2)" --run-level full_model
+          - pytest -s -v tests/e2e/accuracy/hunyuanvideo15_i2v/test_hunyuanvideo15_i2v_video_similarity.py -m "full_model and H100 and B200 and (cards_1 or cards_2)" --run-level full_model
 
       - label: ":full_moon: Diffusion X2V · Perf Test · Single-GPU"
+        source_file_dependencies: diffusion_wan22_perf
         key: nightly-diffusion-x2v-performance-single
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/dfx/perf/results/*.json
         commands:
-          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_wan22_i2v_vllm_omni.json -m "H100 and cards_1"
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
-            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
-            exit $$EXIT
-        mirror_hardwares: h100_1
+          - export BENCHMARK_DIR=tests/dfx/perf/results
+          - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_wan22_i2v_vllm_omni.json -m "H100 and B200 and cards_1"
 
       - label: ":full_moon: Diffusion X2V · Perf Test · 2-GPU"
+        source_file_dependencies:
+          - diffusion_wan22_perf
+          - diffusion_cosmos3_perf
         key: nightly-diffusion-x2v-performance-2gpu
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/dfx/perf/results/*.json
         commands:
-          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
-          - export DIFFUSION_ATTENTION_BACKEND=FLASH_ATTN
+          - export BENCHMARK_DIR=tests/dfx/perf/results
           - |
             set +e
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_wan22_i2v_vllm_omni.json -m "H100 and cards_2"
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_wan22_i2v_vllm_omni.json -m "H100 and B200 and cards_2"
             EXIT1=$$?
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_cosmos3_vllm_omni.json -m "H100 and cards_2"
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_cosmos3_vllm_omni.json -m "H100 and B200 and cards_2"
             EXIT2=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
-            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
             if [ $$EXIT1 -ne 0 ]; then exit $$EXIT1; fi
             exit $$EXIT2
-        mirror_hardwares: h100_2
 
       - label: ":full_moon: Diffusion X2V · Perf Test · HunyuanVideo-1.5"
+        source_file_dependencies: diffusion_hunyuan_video_perf
         key: nightly-diffusion-x2v-performance-hunyuanvideo15
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/dfx/perf/results/*.json
         commands:
-          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
+          - export BENCHMARK_DIR=tests/dfx/perf/results
           - |
             set +e
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_hunyuanvideo15_t2v_vllm_omni.json
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_hunyuanvideo15_t2v_vllm_omni.json -m "H100 and B200 and (cards_1 or cards_2)"
             EXIT1=$$?
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_hunyuanvideo15_i2v_vllm_omni.json
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_hunyuanvideo15_i2v_vllm_omni.json -m "H100 and B200 and (cards_1 or cards_2)"
             EXIT2=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
-            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
             if [ $$EXIT1 -ne 0 ]; then exit $$EXIT1; fi
             exit $$EXIT2
-        mirror_hardwares: h100_2
 
       - label: ":full_moon: Diffusion X2V · Perf Test · MiniMax-H3"
+        source_file_dependencies: diffusion_minimax_h3_perf
         key: nightly-diffusion-x2v-performance-minimax-h3
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/dfx/perf/results/*.json
         commands:
-          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_minimax_h3_vllm_omni.json
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
-            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
-            exit $$EXIT
-        mirror_hardwares: h100_4
+          - export BENCHMARK_DIR=tests/dfx/perf/results
+          - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_minimax_h3_vllm_omni.json -m "H100 and B200 and cards_4"
+
+      - label: ":full_moon: Diffusion X2V · Perf Test · LingBot"
+        source_file_dependencies: diffusion_lingbot_perf
+        key: nightly-diffusion-x2v-performance-lingbot
+        timeout_in_minutes: 180
+        artifact_paths:
+          - tests/dfx/perf/results/*.json
+        commands:
+          - export BENCHMARK_DIR=tests/dfx/perf/results
+          - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_lingbot_video_vllm_omni.json -m "H100 and B200 and cards_1"
 
   - group: ":card_index_dividers: Diffusion Test"
     key: nightly-diffusion-test-group
@@ -450,59 +470,59 @@ steps:
     if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test"
     steps:
       - label: ":full_moon: Diffusion · Distributed Test with L4 · 2-GPU"
+        source_file_dependencies: diffusion_distributed_attention
         timeout_in_minutes: 60
         commands:
-          - pytest -sv tests/diffusion/distributed/ -m 'full_model and cuda and L4 and cards_2' --run-level "full_model"
-        mirror_hardwares: l4_2
+          - pytest -sv tests/diffusion/distributed/ -m 'full_model and cuda and L4 and B200 and cards_2' --run-level "full_model"
 
       - label: ":full_moon: Diffusion · Distributed Test with L4 · 3-GPU"
+        source_file_dependencies: diffusion_distributed_attention
         timeout_in_minutes: 60
         commands:
-          - pytest -sv tests/diffusion/distributed/ -m 'full_model and cuda and L4 and cards_3' --run-level "full_model"
-        mirror_hardwares: l4_3
+          - pytest -sv tests/diffusion/distributed/ -m 'full_model and cuda and L4 and B200 and cards_3' --run-level "full_model"
 
       - label: ":full_moon: Diffusion · Distributed Test with L4 · 4-GPU"
+        source_file_dependencies: diffusion_distributed_attention
         timeout_in_minutes: 60
         commands:
-          - pytest -sv tests/diffusion/distributed/ -m 'full_model and cuda and L4 and cards_4' --run-level "full_model"
-        mirror_hardwares: l4_4
+          - pytest -sv tests/diffusion/distributed/ -m 'full_model and cuda and L4 and B200 and cards_4' --run-level "full_model"
 
       - label: ":full_moon: Diffusion · Distributed Test with H100 · 2-GPU"
+        source_file_dependencies: diffusion_distributed_attention
         timeout_in_minutes: 60
         commands:
-          - pytest -sv tests/diffusion/distributed/ -m 'full_model and cuda and H100 and cards_2' --run-level "full_model"
-        mirror_hardwares: h100_2
+          - pytest -sv tests/diffusion/distributed/ -m 'full_model and cuda and H100 and B200 and cards_2' --run-level "full_model"
 
       - label: ":full_moon: Diffusion · Distributed Test with H100 · 4-GPU"
+        source_file_dependencies: diffusion_distributed_attention
         timeout_in_minutes: 60
         commands:
-          - pytest -sv tests/diffusion/distributed/ -m 'full_model and cuda and H100 and cards_4' --run-level "full_model"
-        mirror_hardwares: h100_4
+          - pytest -sv tests/diffusion/distributed/ -m 'full_model and cuda and H100 and B200 and cards_4' --run-level "full_model"
 
       - label: ":full_moon: Diffusion · Offloader Test"
+        source_file_dependencies: diffusion_offloader
         timeout_in_minutes: 60
         commands:
-          - pytest -sv tests/diffusion/offloader -m "full_model and cuda"
-        mirror_hardwares: l4_1
+          - pytest -sv tests/diffusion/offloader -m "full_model and L4 and B200 and cuda and cards_1"
 
       - label: ":full_moon: Diffusion · Quantization Test with H100 · Single-GPU"
+        source_file_dependencies: diffusion_quantization
         timeout_in_minutes: 60
         commands:
-          - pytest -sv tests/diffusion/quantization -m 'full_model and cuda and H100' --run-level "full_model"
-        mirror_hardwares: h100_1
+          - pytest -sv tests/diffusion/quantization -m 'full_model and cuda and H100 and B200 and cards_1' --run-level "full_model"
 
       - label: ":full_moon: Diffusion · MiniMax H3 FP8 Quality Test with H100 · 2-GPU"
+        source_file_dependencies: diffusion_minimax_h3_function
         timeout_in_minutes: 60
         commands:
-          - pytest -sv tests/diffusion/models/minimax_h3/test_minimax_h3_quantization_quality.py -m 'full_model and cuda and H100' --run-level "full_model"
-        mirror_hardwares: h100_2
+          - pytest -sv tests/diffusion/models/minimax_h3/test_minimax_h3_quantization_quality.py -m 'full_model and cuda and H100 and B200 and cards_2' --run-level "full_model"
 
       - label: ":full_moon: Diffusion · Quantization Test with L4"
+        source_file_dependencies: diffusion_quantization
         timeout_in_minutes: 60
         commands:
           - pip install "vllm-gguf-plugin==0.0.4"
-          - pytest -sv tests/diffusion/quantization -m 'full_model and cuda and L4' --run-level "full_model"
-        mirror_hardwares: l4_1
+          - pytest -sv tests/diffusion/quantization -m 'full_model and cuda and L4 and B200 and cards_1' --run-level "full_model"
 
   # NOTE: ":email: Nightly Collection & Email" is deprecated and will be removed soon.
   # Do not add new depends_on entries or extend this step.
@@ -518,6 +538,9 @@ steps:
       - nightly-diffusion-x2v-performance-single
       - nightly-diffusion-x2v-performance-2gpu
     if: build.env("NIGHTLY") == "1" && build.env("EMAIL_DISTRIBUTION") == "1"
+    artifact_paths:
+      - tests/dfx/perf/results/*.xlsx
+      - tests/dfx/perf/results/*.html
     commands:
       - pip install openpyxl
       - export DEFAULT_INPUT_DIR=tests/dfx/perf/results
@@ -533,7 +556,5 @@ steps:
       - python tools/nightly/generate_nightly_perf_excel.py
       - python tools/nightly/generate_nightly_perf_html.py
       - python tools/nightly/send_nightly_email.py --report-file "tests/dfx/perf/results/*.xlsx, tests/dfx/perf/results/*.html"
-      - buildkite-agent artifact upload "tests/dfx/perf/results/*.xlsx"
-      - buildkite-agent artifact upload "tests/dfx/perf/results/*.html"
     agents:
       queue: "cpu_queue_premerge"

.buildkite/cuda/test-ready.yml

@@ -16,8 +16,27 @@ steps:
           - "pytest -sv tests/model_executor -m 'core_model and cpu'"
         mirror_hardwares: l4_1
 
+      - label: "Simple · CosyVoice3 Packed Flow Test"
+        timeout_in_minutes: 15
+        source_file_dependencies:
+          - tests/model_executor/models/cosyvoice3/test_cosyvoice3_full_batch.py
+          - vllm_omni/model_executor/models/cosyvoice3/
+        commands:
+          - "pytest -sv tests/model_executor/models/cosyvoice3/test_cosyvoice3_full_batch.py -k packed -m 'core_model and cuda' --run-level=core_model"
+        mirror_hardwares: h100_1
+
+      - label: "Simple · MRV2 FA3 Graph Bounds Test"
+        timeout_in_minutes: 15
+        source_file_dependencies:
+          - tests/worker_v2/test_capture_query_bound_gpu.py
+          - tests/worker_v2/test_decode_query_bound_gpu.py
+          - vllm_omni/worker_v2/model_states/omni_model_state.py
+        commands:
+          - "pytest -sv tests/worker_v2/test_capture_query_bound_gpu.py tests/worker_v2/test_decode_query_bound_gpu.py -m 'core_model and cuda and H100' --run-level=core_model"
+        mirror_hardwares: h100_1
+
       - label: "Simple · Diffusion Test"
-        timeout_in_minutes: 20
+        timeout_in_minutes: 30
         commands:
           - "pytest -sv tests/diffusion -m 'core_model and cpu'"
         mirror_hardwares: l4_1
@@ -38,6 +57,12 @@ steps:
     depends_on: upload-ready-pipeline
     steps:
 
+      - label: "Diffusion · FlashAttention Test"
+        timeout_in_minutes: 20
+        commands:
+          - timeout 15m pytest -s -v tests/diffusion/attention/ -m 'core_model and cuda and L4 and cards_1' --run-level "core_model"
+        mirror_hardwares: l4_1
+
       - label: "Diffusion · Offloader Test"
         commands:
           - timeout 40m pytest -s -v tests/diffusion/offloader -m "core_model and cuda" --run-level "core_model"
@@ -54,16 +79,20 @@ steps:
         mirror_hardwares: l4_1
 
       - label: "Diffusion · Distributed Test · Single-GPU"
+        source_file_dependencies: diffusion_distributed_attention
         commands:
           - timeout 20m pytest -sv tests/diffusion/distributed -m 'core_model and cuda and L4 and cards_1' --run-level "core_model"
         mirror_hardwares: l4_1
 
       - label: "Diffusion · Distributed Test · 2-GPU"
+        source_file_dependencies: diffusion_distributed_attention
         commands:
           - timeout 20m pytest -sv tests/diffusion/distributed -m 'core_model and cuda and L4 and cards_2' --run-level "core_model"
+          - timeout 10m pytest -sv tests/diffusion/attention/test_ulysses_uaa.py -m 'core_model and cuda and L4 and cards_2' --run-level "core_model"
         mirror_hardwares: l4_2
 
       - label: "Diffusion · Distributed & Attention Test · 4-GPU"
+        source_file_dependencies: diffusion_distributed_attention
         commands:
           - timeout 20m pytest -sv tests/diffusion/distributed -m 'core_model and cuda and L4 and cards_4' --run-level "core_model"
           - timeout 20m pytest -sv tests/diffusion/attention/test_attention_sp.py -m 'core_model and cuda and L4 and cards_4' --run-level "core_model"
@@ -76,7 +105,7 @@ steps:
 
       - label: "Diffusion · Other Test"
         commands:
-          - timeout 15m pytest -sv tests/diffusion/*.py tests/diffusion/ar_diffusion -m 'core_model and cuda' --run-level "core_model"
+          - timeout 15m pytest -sv tests/diffusion/*.py tests/diffusion/ar_diffusion tests/diffusion/layers/test_adalayernorm_*.py -m 'core_model and cuda' --run-level "core_model"
         mirror_hardwares: l4_1
 
   - group: ":card_index_dividers: Tiny Model Tests [Single GPU]"
@@ -93,7 +122,7 @@ steps:
     commands:
       - |
         timeout 15m bash -c "
-                  pytest -sv tests/engine/ tests/model_executor/ tests/worker/test_batched_omni_output.py -m 'core_model and cuda' --run-level 'core_model'
+                  pytest -sv tests/engine/ tests/model_executor/ tests/worker/test_batched_omni_output.py tests/worker/test_gpu_generation_async_output.py tests/worker_v2/ -m 'core_model and cuda' --run-level 'core_model'
         "
     mirror_hardwares: l4_1
 
@@ -108,11 +137,18 @@ steps:
 
   - label: "Entrypoints Test"
     depends_on: upload-ready-pipeline
+    # Workaround https://github.com/vllm-project/vllm-omni/issues/8016: shutdown()
+    # leaves inline UniProc DiffusionWorker VRAM mapped in the pytest parent.
+    # Isolate that leftover case in its own process so the driver drops the
+    # mapping; the rest stays a directory sweep with --deselect. Revert to a
+    # single sweep after the product shutdown fix.
     commands:
       - |
         timeout 30m bash -c "
+          set -eu
           export VLLM_WORKER_MULTIPROC_METHOD=spawn
-          pytest -sv tests/entrypoints/ -m 'core_model and cuda' --run-level 'core_model'
+          pytest -sv tests/entrypoints/test_async_omni.py::test_diffusion_generate_request_id -m 'core_model and cuda' --run-level 'core_model'
+          pytest -sv tests/entrypoints/ -m 'core_model and cuda' --run-level 'core_model' --deselect tests/entrypoints/test_async_omni.py::test_diffusion_generate_request_id
         "
     mirror_hardwares: l4_1
 
@@ -120,29 +156,14 @@ steps:
     depends_on: upload-ready-pipeline
     steps:
       - label: "Custom Pipeline Test"
-        source_file_dependencies:
-          - tests/e2e/features/custom_pipeline/
-          - tests/e2e/features/helpers/custom_pipeline.py
-          - vllm_omni/diffusion/
-          - vllm_omni/engine/
-          - vllm_omni/entrypoints/
-          - vllm_omni/inputs/
-          - vllm_omni/outputs/
+        source_file_dependencies: custom_pipeline
         commands:
           - timeout 20m pytest -s -v tests/e2e/features/custom_pipeline/ -m "core_model"
         mirror_hardwares: l4_1
 
 
       - label: "RLHF · VeRL-Omni E2E Test"
-        source_file_dependencies:
-          - tests/e2e/features/rlhf_test/test_verl_omni_e2e.py
-          - tests/e2e/features/helpers/verl_omni_server.py
-          - tests/e2e/features/helpers/custom_pipeline.py
-          - vllm_omni/diffusion/
-          - vllm_omni/engine/
-          - vllm_omni/entrypoints/
-          - vllm_omni/inputs/
-          - vllm_omni/outputs/
+        source_file_dependencies: rlhf_verl_omni
         commands:
           - |
             timeout 20m bash -c "
@@ -151,17 +172,7 @@ steps:
         mirror_hardwares: l4_1
 
       - label: "Omni · Qwen3-Omni Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_qwen3_omni.py
-          - tests/helpers/runtime.py
-          - vllm_omni/entrypoints/openai/serving_chat.py
-          - vllm_omni/model_executor/models/qwen3_omni/
-          - vllm_omni/model_executor/models/common/qwen3_code_predictor.py
-          - vllm_omni/model_executor/models/common/snake_activation.py
-          - vllm_omni/model_executor/models/qwen2_5_omni/qwen2_5_omni_thinker.py
-          - vllm_omni/model_executor/models/qwen3_tts/cuda_graph_decoder_wrapper.py
-          - vllm_omni/model_executor/stage_input_processors/qwen3_omni.py
-          - vllm_omni/deploy/qwen3_omni_moe.yaml
+        source_file_dependencies: omni_qwen3_omni_function
         commands:
           - |
             timeout 20m bash -c "
@@ -170,14 +181,7 @@ steps:
         mirror_hardwares: h100_2
 
       - label: "Omni · MiniCPM-o 4.5 Offline Test"
-        source_file_dependencies:
-          - tests/e2e/offline_inference/test_minicpmo_4_5.py
-          - vllm_omni/model_executor/models/minicpmo_4_5/
-          - vllm_omni/model_executor/models/step_audio2/
-          - vllm_omni/model_executor/models/cosyvoice3/code2wav_core/
-          - vllm_omni/model_executor/models/cosyvoice3/utils.py
-          - vllm_omni/model_executor/stage_input_processors/minicpmo_4_5_omni.py
-          - vllm_omni/deploy/minicpmo_4_5.yaml
+        source_file_dependencies: omni_minicpmo_4_5_function
         commands:
           - |
             timeout 20m bash -c "
@@ -186,15 +190,7 @@ steps:
         mirror_hardwares: h100_1
 
       - label: "Omni · MiniCPM-o 4.5 Online Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_minicpmo_4_5.py
-          - vllm_omni/transformers_utils/chat_templates/minicpmo45_native.jinja
-          - vllm_omni/model_executor/models/minicpmo_4_5/
-          - vllm_omni/model_executor/models/step_audio2/
-          - vllm_omni/model_executor/models/cosyvoice3/code2wav_core/
-          - vllm_omni/model_executor/models/cosyvoice3/utils.py
-          - vllm_omni/model_executor/stage_input_processors/minicpmo_4_5_omni.py
-          - vllm_omni/deploy/minicpmo_4_5.yaml
+        source_file_dependencies: omni_minicpmo_4_5_function
         commands:
           - |
             timeout 20m bash -c "
@@ -203,48 +199,7 @@ steps:
         mirror_hardwares: h100_1
 
       - label: "Omni · MiniCPM-o 4.5 Duplex Test"
-        source_file_dependencies:
-          - tests/assets/minicpmo_4_5/response_required_16k.wav
-          - tests/assets/minicpmo_4_5/soft_interrupt_16k.wav
-          - tests/e2e/online_serving/test_minicpmo_4_5_duplex.py
-          - tests/e2e/online_serving/test_minicpmo_4_5_duplex_expansion.py
-          - tests/e2e/online_serving/test_minicpmo_realtime_duplex_drivers.py
-          - tests/e2e/online_serving/helpers/minicpmo_realtime_duplex_scenarios.py
-          - tests/e2e/online_serving/run_minicpmo_realtime_duplex_multi_session.py
-          - tests/e2e/online_serving/run_minicpmo_realtime_duplex_server_vad.py
-          - tests/e2e/online_serving/run_minicpmo_realtime_duplex_soft_interrupt.py
-          - vllm_omni/deploy/minicpmo_4_5.yaml
-          - tests/e2e/online_serving/helpers/minicpmo_4_5_duplex.py
-          - vllm_omni/engine/duplex/
-          - vllm_omni/entrypoints/duplex/
-          # Public duplex client library + its unit suites: client-side
-          # changes must re-run the duplex e2e group, which includes the
-          # live client probe (test_duplex_client_live.py, single GPU).
-          - vllm_omni/clients/
-          - tests/clients/
-          - tests/e2e/online_serving/test_duplex_client_live.py
-          - vllm_omni/entrypoints/duplex_request_client.py
-          - vllm_omni/outputs/duplex.py
-          - vllm_omni/model_executor/duplex_sampling.py
-          - tests/engine/duplex/
-          - tests/entrypoints/duplex/
-          - tests/model_executor/models/minicpmo_4_5/duplex/
-          # Legacy Realtime probe client + video stacking used by the e2e drivers.
-          - vllm_omni/experimental/fullduplex/
-          - vllm_omni/model_executor/models/minicpmo_4_5/
-          - vllm_omni/model_executor/models/step_audio2/
-          - vllm_omni/model_executor/models/cosyvoice3/code2wav_core/
-          - vllm_omni/model_executor/models/cosyvoice3/utils.py
-          - vllm_omni/model_executor/stage_input_processors/minicpmo_4_5_omni.py
-          - vllm_omni/core/sched/
-          - vllm_omni/outputs/multimodal_accumulation.py
-          - vllm_omni/engine/stage_init_utils.py
-          - vllm_omni/engine/async_omni_engine.py
-          - vllm_omni/engine/orchestrator.py
-          - vllm_omni/entrypoints/cli/serve.py
-          - vllm_omni/entrypoints/openai/api_server.py
-          - vllm_omni/entrypoints/openai/duplex_capability.py
-          - vllm_omni/config/stage_config.py
+        source_file_dependencies: omni_minicpmo_4_5_duplex_function
         commands:
           - |
             timeout 35m bash -c "
@@ -254,29 +209,24 @@ steps:
 
       - label: "Omni · MiniCPM-o 4.5 Perf Test"
         timeout_in_minutes: 60
-        source_file_dependencies:
-          - tests/dfx/perf/tests/test_minicpmo_4_5_ready.json
-          - vllm_omni/deploy/minicpmo_4_5.yaml
-          - vllm_omni/model_executor/models/minicpmo_4_5/
-          - vllm_omni/model_executor/stage_input_processors/minicpmo_4_5_omni.py
+        source_file_dependencies: omni_minicpmo_4_5_perf
         commands:
           - export BENCHMARK_DIR=tests/dfx/perf/results
           - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_minicpmo_4_5_ready.json -m 'core_model and cuda'
         mirror_hardwares: h100_1
 
-      - label: "TTS · Qwen3-TTS CustomVoice Test"
+      - label: "TTS · Higgs v3 Full Response Test"
         source_file_dependencies:
-          - tests/helpers/media.py
-          - tests/e2e/online_serving/test_qwen3_tts_customvoice.py
-          - vllm_omni/model_executor/models/qwen3_tts/
-          - vllm_omni/model_executor/models/common/qwen3_code_predictor.py
-          - vllm_omni/model_executor/models/common/snake_activation.py
-          - vllm_omni/model_executor/models/common/alias_free_activation.py
-          - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
-          - vllm_omni/deploy/qwen3_tts.yaml
-          - vllm_omni/entrypoints/openai/serving_speech.py
-          - vllm_omni/entrypoints/openai/tts_adapters/
-          - vllm_omni/engine/stage_pool.py
+          - tests/e2e/online_serving/test_higgs_audio_v3.py
+          - vllm_omni/model_executor/models/higgs_audio_v3/
+          - vllm_omni/model_executor/stage_input_processors/higgs_audio_v3.py
+          - vllm_omni/deploy/higgs_multimodal_qwen3.yaml
+        commands:
+          - timeout 20m pytest -s -v tests/e2e/online_serving/test_higgs_audio_v3.py::test_full_response_v1_preserves_audio -m 'core_model and cuda' --run-level 'core_model'
+        mirror_hardwares: h100_1
+
+      - label: "TTS · Qwen3-TTS CustomVoice Test"
+        source_file_dependencies: tts_qwen3_tts_function
         commands:
           - |
             timeout 20m bash -c "
@@ -286,25 +236,17 @@ steps:
         mirror_hardwares: l4_1
 
       - label: "Diffusion · Wan22 Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_wan22_t2v.py
-          - vllm_omni/diffusion/models/wan2_2/
-          - vllm_omni/diffusion/models/dmd2/
+        source_file_dependencies: diffusion_wan22_function
         commands:
           - |
             timeout 40m bash -c "
               export VLLM_IMAGE_FETCH_TIMEOUT=60
-              pytest -s -v tests/e2e/online_serving/test_wan22_t2v.py -m 'core_model' --run-level 'core_model'
+              pytest -s -v tests/e2e/online_serving/test_wan22_t2v.py -m 'core_model and cuda' --run-level 'core_model'
             "
         mirror_hardwares: h100_1
 
       - label: "Diffusion · Cosmos3 Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_cosmos3.py
-          - vllm_omni/diffusion/models/cosmos3/
-          - vllm_omni/diffusion/distributed/autoencoders/autoencoder_kl_wan.py
-          - vllm_omni/experimental/world_models/
-          - vllm_omni/model_extras/cosmos3.py
+        source_file_dependencies: diffusion_cosmos3_function
         commands:
           - |
             timeout 40m bash -c "
@@ -315,14 +257,7 @@ steps:
 
       - label: "Diffusion · MiniMax-H3 DLO DP2 T2VA Test"
         timeout_in_minutes: 90
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_minimax_h3_dlo_dp2_t2va.py
-          - tests/helpers/assertions.py
-          - tests/helpers/runtime.py
-          - vllm_omni/diffusion/models/minimax_h3/
-          - vllm_omni/diffusion/offloader/
-          - vllm_omni/diffusion/distributed/
-          - vllm_omni/entrypoints/openai/
+        source_file_dependencies: diffusion_minimax_h3_function
         commands:
           - |
             timeout 80m bash -c "
@@ -332,11 +267,7 @@ steps:
         mirror_hardwares: h100_2
 
       - label: "Diffusion · Qwen Image Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_qwen_image.py
-          - vllm_omni/diffusion/models/qwen_image/
-          - vllm_omni/diffusion/models/dmd2/
-          - vllm_omni/diffusion/distributed/autoencoders/autoencoder_kl_qwenimage.py
+        source_file_dependencies: diffusion_qwen_image_function
         commands:
           - |
             timeout 40m bash -c "
@@ -346,9 +277,7 @@ steps:
         mirror_hardwares: h100_1
 
       - label: "Diffusion · HunyuanVideo-1.5 Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_hunyuan_video_15_expansion.py
-          - vllm_omni/diffusion/models/hunyuan_video/
+        source_file_dependencies: diffusion_hunyuan_video_function
         commands:
           - |
             timeout 40m bash -c "
@@ -356,3 +285,13 @@ steps:
               pytest -s -v tests/e2e/online_serving/test_hunyuan_video_15_expansion.py -m 'core_model and diffusion' --run-level 'core_model'
             "
         mirror_hardwares: h100_1
+
+      - label: "Diffusion · LingBot-Video Test"
+        source_file_dependencies: diffusion_lingbot_function
+        commands:
+          - |
+            timeout 40m bash -c "
+              export VLLM_IMAGE_FETCH_TIMEOUT=60
+              pytest -s -v tests/e2e/online_serving/test_lingbot_video.py -m 'core_model and cuda' --run-level 'core_model'
+            "
+        mirror_hardwares: h100_1

.buildkite/cuda/test-weekly.yml

@@ -12,6 +12,8 @@ steps:
     depends_on: upload-weekly-pipeline
     if: build.env("WEEKLY") == "1"
     timeout_in_minutes: 60
+    artifact_paths:
+      - coverage-core-model-cpu-*.xml.gz
     commands:
       - |
         set +e
@@ -19,7 +21,6 @@ steps:
         pytest -sv tests/ -m 'core_model and cpu' --cov=vllm_omni --cov-report=term-missing:skip-covered --cov-report=xml:$${REPORT}
         EXIT=$$?
         gzip -9 -f "$${REPORT}"
-        buildkite-agent artifact upload "$${REPORT}.gz"
         exit $$EXIT
     mirror_hardwares: l4_1
 
@@ -29,36 +30,42 @@ steps:
     steps:
 
       - label: "Reliability Test · qwen3-omni"
+        source_file_dependencies: omni_qwen3_omni_reliability
         timeout_in_minutes: 180
         commands:
           - pytest -s -v tests/dfx/reliability/test_reliability_qwen3_omni.py -m "slow"
         mirror_hardwares: h100_2
 
       - label: "Reliability Test · wan22"
+        source_file_dependencies: diffusion_wan22_reliability
         timeout_in_minutes: 180
         commands:
           - pytest -s -v tests/dfx/reliability/test_reliability_wan22.py -m "slow"
         mirror_hardwares: h100_2
 
       - label: "Reliability Test · hunyuan-image"
+        source_file_dependencies: diffusion_hunyuan_image3_reliability
         timeout_in_minutes: 180
         commands:
           - pytest -s -v tests/dfx/reliability/test_reliability_hunyuan_image.py -m "slow"
         mirror_hardwares: h100_4
 
       - label: "Reliability Test · Invalid parameters · H100 · Single-GPU"
+        source_file_dependencies: reliability_invalid_param_h100
         timeout_in_minutes: 90
         commands:
           - pytest -s -v tests/dfx/reliability/invalid_param_test/ -m "slow and H100 and cards_1"
         mirror_hardwares: h100_1
 
       - label: "Reliability Test · Invalid parameters · H100 · 2-GPU"
+        source_file_dependencies: reliability_invalid_param_h100
         timeout_in_minutes: 90
         commands:
           - pytest -s -v tests/dfx/reliability/invalid_param_test/ -m "slow and H100 and cards_2"
         mirror_hardwares: h100_2
 
       - label: "Reliability Test · Invalid parameters · L4"
+        source_file_dependencies: reliability_invalid_param_l4
         timeout_in_minutes: 120
         commands:
           - pytest -s -v tests/dfx/reliability/invalid_param_test/ -m "slow and L4"
@@ -70,42 +77,35 @@ steps:
     steps:
 
       - label: ":full_moon: Omni · Perf Test · vLLM Text"
+        source_file_dependencies: omni_qwen3_omni_perf
         key: weekly-omni-performance-vllm-text
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/dfx/perf/results/*.json
         commands:
           - export BENCHMARK_DIR=tests/dfx/perf/results
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen3_omni_vllm_text.json
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
-            exit $$EXIT
+          - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen3_omni_vllm_text.json
         mirror_hardwares: h100_2
 
       - label: ":full_moon: Omni · Perf Test · Async Chunk · Random"
         key: weekly-omni-performance-async-chunk-random
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/dfx/perf/results/*.json
         commands:
           - export BENCHMARK_DIR=tests/dfx/perf/results
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen3_omni_async_chunk.json -m "H100 and slow"
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
-            exit $$EXIT
+          - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen3_omni_async_chunk.json -m "H100 and slow"
         mirror_hardwares: h100_2
 
       - label: ":full_moon: Omni · Perf Test · Multi-Replica"
+        source_file_dependencies: omni_qwen3_omni_perf
         key: weekly-omni-performance-multi-replicas
         timeout_in_minutes: 180
+        artifact_paths:
+          - tests/dfx/perf/results/*.json
         commands:
           - export BENCHMARK_DIR=tests/dfx/perf/results
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen3_omni_multi_replicas.json
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
-            exit $$EXIT
+          - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen3_omni_multi_replicas.json
         mirror_hardwares: h100_3
 
   - group: ":card_index_dividers: E2E Tests"
@@ -152,15 +152,122 @@ steps:
             --run-level "full_model"
         mirror_hardwares: h100_4
 
-      - label: "TTS · L4"
-        timeout_in_minutes: 120
+      - label: "TTS · Moss · L4"
+        timeout_in_minutes: 90
         commands:
           - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
           - export VLLM_USE_DEEP_GEMM="0"
           - export VLLM_MOE_USE_DEEP_GEMM="0"
           - |
             failed=0
-            for f in $(pytest --collect-only -q tests/e2e/ -m "slow and L4 and tts" --run-level "full_model" | grep '\.py::' | sed -E 's|::.*||' | sort -u); do
+            for f in \
+              tests/e2e/offline_inference/test_moss_tts_expansion.py \
+              tests/e2e/offline_inference/test_moss_tts_nano_expansion.py \
+              tests/e2e/offline_inference/test_moss_tts_realtime.py \
+              tests/e2e/offline_inference/test_moss_tts_realtime_expansion.py \
+              tests/e2e/online_serving/test_moss_tts_expansion.py \
+              tests/e2e/online_serving/test_moss_tts_nano_expansion.py
+            do
+              pytest -s -v "$$f" -m "slow and L4 and tts" --run-level "full_model" || failed=1
+            done
+            exit $$failed
+        mirror_hardwares: l4_1
+
+      - label: "TTS · VoxCPM2 · Voxtral · L4"
+        timeout_in_minutes: 90
+        commands:
+          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+          - export VLLM_USE_DEEP_GEMM="0"
+          - export VLLM_MOE_USE_DEEP_GEMM="0"
+          - |
+            failed=0
+            for f in \
+              tests/e2e/offline_inference/test_voxcpm2_tts.py \
+              tests/e2e/online_serving/test_voxcpm2_tts.py \
+              tests/e2e/online_serving/test_voxcpm2_tts_expansion.py \
+              tests/e2e/online_serving/test_voxcpm2_tts_lora.py \
+              tests/e2e/offline_inference/test_voxtral_tts_expansion.py \
+              tests/e2e/online_serving/test_voxtral_tts_expansion.py
+            do
+              pytest -s -v "$$f" -m "slow and L4 and tts" --run-level "full_model" || failed=1
+            done
+            exit $$failed
+        mirror_hardwares: l4_1
+
+      - label: "TTS · Higgs · L4"
+        timeout_in_minutes: 90
+        commands:
+          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+          - export VLLM_USE_DEEP_GEMM="0"
+          - export VLLM_MOE_USE_DEEP_GEMM="0"
+          - |
+            failed=0
+            for f in \
+              tests/e2e/offline_inference/test_higgs_audio_v2_expansion.py \
+              tests/e2e/online_serving/test_higgs_audio_v2_expansion.py \
+              tests/e2e/online_serving/test_higgs_audio_v3.py
+            do
+              pytest -s -v "$$f" -m "slow and L4 and tts" --run-level "full_model" || failed=1
+            done
+            exit $$failed
+        mirror_hardwares: l4_1
+
+      - label: "TTS · OmniVoice · Step-Audio · L4"
+        timeout_in_minutes: 90
+        commands:
+          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+          - export VLLM_USE_DEEP_GEMM="0"
+          - export VLLM_MOE_USE_DEEP_GEMM="0"
+          - |
+            failed=0
+            for f in \
+              tests/e2e/offline_inference/test_omnivoice_expansion.py \
+              tests/e2e/online_serving/test_omnivoice_expansion.py \
+              tests/e2e/online_serving/test_omnivoice_parity.py \
+              tests/e2e/offline_inference/test_step_audio2_expansion.py
+            do
+              pytest -s -v "$$f" -m "slow and L4 and tts" --run-level "full_model" || failed=1
+            done
+            exit $$failed
+        mirror_hardwares: l4_1
+
+      - label: "TTS · Ming · GLM · IndexTTS · Breeze · L4"
+        timeout_in_minutes: 90
+        commands:
+          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+          - export VLLM_USE_DEEP_GEMM="0"
+          - export VLLM_MOE_USE_DEEP_GEMM="0"
+          - |
+            failed=0
+            for f in \
+              tests/e2e/offline_inference/test_ming_tts_expansion.py \
+              tests/e2e/online_serving/test_ming_tts_expansion.py \
+              tests/e2e/offline_inference/test_glm_tts_expansion.py \
+              tests/e2e/online_serving/test_glm_tts_expansion.py \
+              tests/e2e/offline_inference/test_indextts2_expansion.py \
+              tests/e2e/online_serving/test_indextts2_expansion.py \
+              tests/e2e/online_serving/test_breeze_tts_2.py
+            do
+              pytest -s -v "$$f" -m "slow and L4 and tts" --run-level "full_model" || failed=1
+            done
+            exit $$failed
+        mirror_hardwares: l4_1
+
+      - label: "TTS · Audex · CosyVoice · Qwen3-TTS · L4"
+        timeout_in_minutes: 90
+        commands:
+          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+          - export VLLM_USE_DEEP_GEMM="0"
+          - export VLLM_MOE_USE_DEEP_GEMM="0"
+          - |
+            failed=0
+            for f in \
+              tests/e2e/offline_inference/test_audex_tts.py \
+              tests/e2e/offline_inference/test_audex_s2t.py \
+              tests/e2e/offline_inference/test_audex_tta.py \
+              tests/e2e/offline_inference/test_cosyvoice3_expansion.py \
+              tests/e2e/online_serving/test_qwen3_tts_word_timestamps.py
+            do
               pytest -s -v "$$f" -m "slow and L4 and tts" --run-level "full_model" || failed=1
             done
             exit $$failed
@@ -183,13 +290,14 @@ steps:
       # from downgrading transformers; see
       # examples/offline_inference/text_to_speech/README.md.
       - label: "TTS · Gepard-1.0 · L4"
-        timeout_in_minutes: 60
+        timeout_in_minutes: 90
         commands:
           - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
           - printf '%s\n' "transformers>=5.5.3" "huggingface-hub>=1.0" > /tmp/gepard-overrides.txt
           - uv pip install --python "$(python3 -c 'import sys; print(sys.executable)')" "nemo_toolkit[tts]==2.7.3" --overrides /tmp/gepard-overrides.txt
           - >-
             pytest -s -v tests/e2e/offline_inference/test_gepard_tts.py
+            tests/e2e/online_serving/test_gepard_tts.py
             -m "tts and cuda"
             --run-level "full_model"
         mirror_hardwares: l4_1
@@ -221,24 +329,289 @@ steps:
             --run-level "full_model"
         mirror_hardwares: l4_4
 
-      - label: "Diffusion · H100 · Single-GPU"
-        timeout_in_minutes: 150
+      # Split the old shared H100 Single-GPU sweep: one pytest process collected
+      # every slow+cards_1 diffusion file and exceeded 150 minutes.
+      - label: ":robot_face: Robot Policy OpenPI · π0.5 · H100 · Single-GPU"
+        timeout_in_minutes: 120
         commands:
+          - pip install --no-deps "openpi-client @ git+https://github.com/Physical-Intelligence/openpi.git@215abfb217dbac7d5f1273282331b9b1866c0479#subdirectory=packages/openpi-client"
           - >-
             pytest -sv
-            tests/e2e/ -m "slow and diffusion and H100 and cards_1"
+            tests/e2e/online_serving/test_pi05_expansion.py
+            -m "slow and diffusion and H100 and cards_1"
             --run-level "full_model"
         mirror_hardwares: h100_1
 
-      - label: "Diffusion · H100 · 2-GPU"
+      - label: "Diffusion · Qwen-Image · H100 · Single-GPU"
+        timeout_in_minutes: 120
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_qwen_image_edit.py
+            tests/e2e/online_serving/test_qwen_image_edit_expansion.py
+            tests/e2e/online_serving/test_qwen_image_layered.py
+            tests/e2e/online_serving/test_qwen_image_layered_expansion.py
+            tests/e2e/accuracy/test_qwen_image_edit.py
+            tests/e2e/accuracy/test_qwen_image_layered.py
+            -m "slow and diffusion and H100 and cards_1"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · Wan2.1 VACE · H100 · Single-GPU"
         timeout_in_minutes: 90
         commands:
           - >-
             pytest -sv
-            tests/e2e/ -m "slow and diffusion and H100 and cards_2"
+            tests/e2e/online_serving/test_wan_2_1_vace_expansion.py
+            -m "slow and diffusion and H100 and cards_1"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · LongCat · H100 · Single-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_longcat_image_expansion.py
+            tests/e2e/online_serving/test_longcat_image_edit_expansion.py
+            -m "slow and diffusion and H100 and cards_1"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · Bagel · H100 · Single-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_bagel.py
+            tests/e2e/online_serving/test_bagel_expansion.py
+            tests/e2e/offline_inference/test_bagel.py
+            tests/e2e/offline_inference/test_bagel_expansion.py
+            -m "slow and diffusion and H100 and cards_1"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · Boogu · H100 · Single-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_boogu_image.py
+            tests/e2e/online_serving/test_boogu_image_edit.py
+            tests/e2e/online_serving/test_boogu_image_expansion.py
+            tests/e2e/online_serving/test_boogu_image_edit_expansion.py
+            -m "slow and diffusion and H100 and cards_1"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · Flux · H100 · Single-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_flux_2_dev_expansion.py
+            tests/e2e/offline_inference/test_flux1_schnell_expansion.py
+            -m "slow and diffusion and H100 and cards_1"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · LTX · H100 · Single-GPU"
+        timeout_in_minutes: 120
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_ltx.py
+            tests/e2e/online_serving/test_ltx25.py
+            tests/e2e/accuracy/ltx/test_ltx_official_similarity.py
+            tests/e2e/accuracy/ltx/test_ltx25_official_similarity.py
+            -m "slow and diffusion and H100 and cards_1"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · Krea2 · H100 · Single-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_krea2_expansion.py
+            tests/e2e/offline_inference/test_krea2_expansion.py
+            -m "slow and diffusion and H100 and cards_1"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · HiDream-I1 · H100 · Single-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_hidream_i1_full_expansion.py
+            -m "slow and diffusion and H100 and cards_1"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · SANA-Video · H100 · Single-GPU"
+        source_file_dependencies: diffusion_sana_video_function
+        timeout_in_minutes: 180
+        commands:
+          - export SANA_VIDEO_GOLDEN_BASE_URL=https://vllm-public-assets.s3.us-west-2.amazonaws.com/omni-assets/sana-video/v2
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_sana_video_expansion.py
+            tests/e2e/accuracy/sana_video/
+            -m "slow and diffusion and H100 and cards_1"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · SenseNova-U1 · H100 · Single-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/offline_inference/test_sensenova_u1_expansion.py
+            tests/e2e/offline_inference/test_sensenova_u1_text2img_expansion.py
+            tests/e2e/offline_inference/test_sensenova_u1_img2img_expansion.py
+            -m "slow and diffusion and H100 and cards_1"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · MammothModa2 · H100 · Single-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/offline_inference/test_mammoth_moda2_expansion.py
+            -m "slow and diffusion and H100 and cards_1"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · Helios · H100 · Single-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/accuracy/test_video_streaming_output_similarity.py
+            -m "slow and diffusion and H100 and cards_1"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      # Split the old shared H100 2-GPU sweep: one pytest process collected every
+      # slow+cards_2 diffusion file and exceeded 90 minutes.
+      - label: "Diffusion · Qwen-Image · H100 · 2-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_qwen_image_edit_expansion.py
+            tests/e2e/online_serving/test_qwen_image_layered_expansion.py
+            -m "slow and diffusion and H100 and cards_2"
             --run-level "full_model"
         mirror_hardwares: h100_2
 
+      - label: "Diffusion · Wan2.1 VACE · H100 · 2-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_wan_2_1_vace_expansion.py
+            -m "slow and diffusion and H100 and cards_2"
+            --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: "Diffusion · LongCat · H100 · 2-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_longcat_image_expansion.py
+            tests/e2e/online_serving/test_longcat_image_edit_expansion.py
+            -m "slow and diffusion and H100 and cards_2"
+            --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: "Diffusion · Bagel · H100 · 2-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_bagel_expansion.py
+            tests/e2e/offline_inference/test_bagel_expansion.py
+            -m "slow and diffusion and H100 and cards_2"
+            --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: "Diffusion · Boogu · H100 · 2-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_boogu_image_expansion.py
+            tests/e2e/online_serving/test_boogu_image_edit_expansion.py
+            -m "slow and diffusion and H100 and cards_2"
+            --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: "Diffusion · Flux · H100 · 2-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_flux_2_dev_expansion.py
+            tests/e2e/offline_inference/test_flux1_kontext_expansion.py
+            -m "slow and diffusion and H100 and cards_2"
+            --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: "Diffusion · GLM-Image · H100 · 2-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_glm_image_expansion.py
+            tests/e2e/offline_inference/test_glm_image_autoround_w4a16_expansion.py
+            -m "slow and diffusion and H100 and cards_2"
+            --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: "Diffusion · LTX · H100 · 2-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_ltx.py
+            -m "slow and diffusion and H100 and cards_2"
+            --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: "Diffusion · Krea2 · H100 · 2-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/offline_inference/test_krea2_expansion.py
+            -m "slow and diffusion and H100 and cards_2"
+            --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: "Diffusion · DreamZero · H100 · 2-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_dreamzero_expansion.py
+            tests/e2e/accuracy/test_dreamzero.py
+            -m "slow and diffusion and H100 and cards_2"
+            --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: "Diffusion · H100 · 3-GPU"
+        timeout_in_minutes: 120
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/ -m "slow and diffusion and H100 and cards_3"
+            --run-level "full_model"
+        mirror_hardwares: h100_3
+
       - label: "Diffusion · H100 · 4-GPU"
         timeout_in_minutes: 90
         commands:
@@ -246,4 +619,33 @@ steps:
             pytest -sv
             tests/e2e/ -m "slow and diffusion and H100 and cards_4"
             --run-level "full_model"
+            --ignore=tests/e2e/online_serving/test_magi2.py
+        mirror_hardwares: h100_4
+
+      # Dedicated: HiDream-O1 accuracy needs a long TP2 run; the shared H100
+      # 2-GPU sweep is only 90 minutes.
+      - label: "Diffusion · HiDream-O1-Image Accuracy · H100 · 2-GPU"
+        timeout_in_minutes: 180
+        commands:
+          - pytest -s -v tests/e2e/accuracy/test_hidream_o1_image.py -m "slow and diffusion and cuda and H100 and cards_2" --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      # Dedicated: Boogu CFG-parallel size 3 needs 3 GPUs; weekly has no H100 3-GPU sweep.
+      - label: "Diffusion · Boogu Edit CFG · H100 · 3-GPU"
+        timeout_in_minutes: 120
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_boogu_image_edit_expansion.py
+            -k "test_double_guidance_cfg_parallel and cfg_parallel_3_full"
+            -m "slow and diffusion and cuda and H100 and cards_3"
+            --run-level "full_model"
+        mirror_hardwares: h100_3
+
+      # Dedicated: MAGI-2 Preview serving smoke needs ~30 min server init; the
+      # shared H100 4-GPU sweep is only 90 minutes.
+      - label: "Diffusion · MAGI-2 Preview Serving Smoke · H100 · 4-GPU"
+        timeout_in_minutes: 180
+        commands:
+          - pytest -s -v tests/e2e/online_serving/test_magi2.py -m "slow and diffusion and cuda and H100 and cards_4" --run-level "full_model"
         mirror_hardwares: h100_4

.buildkite/intel/pipeline-intel.yml

@@ -10,7 +10,7 @@ steps:
           DOCKER_BUILDKIT: "1"
           # Buildkite will automatically replace this with the actual commit hash
           VLLM_IMAGE_TAG: "${BUILDKITE_COMMIT}"
-          VLLM_VERSION: "v0.28.0"
+          VLLM_VERSION: "v0.31.0"
         priority: 100
         timeout_in_minutes: 60
         soft_fail: false

.buildkite/npu/bootstrap-upload-steps.yml

@@ -93,7 +93,7 @@ steps:
     command: |
       python3 .buildkite/common/scripts/upload_pipeline.py --upload .buildkite/npu/test-npu-ready.yml
 
-  # L4 Test — main+NIGHTLY=1 (scheduled), or PR with specific label (e.g. add label then Rebuild)
+  # L4 Test — main+NIGHTLY=1 (scheduled), or PR nightly-test label
   - label: "Upload Nightly Pipeline"
     depends_on:
       - image-build-a2
@@ -102,6 +102,11 @@ steps:
       - image-build-310p
     key: upload-nightly-pipeline
     commands:
-      - python3 .buildkite/common/scripts/upload_pipeline.py --upload .buildkite/npu/test-npu-nightly.yml
+      - |
+        EXTRA=""
+        if [ "${NIGHTLY:-}" = "1" ] && [ "${BUILDKITE_BRANCH}" = "main" ]; then
+          EXTRA="--all"
+        fi
+        python3 .buildkite/common/scripts/upload_pipeline.py --upload $$EXTRA .buildkite/npu/test-npu-nightly.yml
     agents:
       queue: "cpu_queue_premerge"

.buildkite/npu/pipeline-npu-a3.yml

@@ -96,13 +96,7 @@
   #   key: upload-nightly-pipeline
   #   if: >-
   #     (build.branch == "main" && build.env("NIGHTLY") == "1") ||
-  #     (build.branch != "main" && (
-  #       build.pull_request.labels includes "nightly-test" ||
-  #       build.pull_request.labels includes "omni-test" ||
-  #       build.pull_request.labels includes "tts-test" ||
-  #       build.pull_request.labels includes "diffusion-x2iat-test" ||
-  #       build.pull_request.labels includes "diffusion-x2v-test"
-  #     ))
+  #     (build.branch != "main" && build.pull_request.labels includes "nightly-test")
   #   commands:
   #     - buildkite-agent pipeline upload .buildkite/npu/test-npu-nightly.yml
   #   agents:

.buildkite/npu/test-npu-nightly.yml

@@ -4,10 +4,10 @@ steps:
     depends_on: upload-nightly-pipeline
     if: >-
       build.env("NIGHTLY") == "1" ||
-      build.pull_request.labels includes "nightly-test" ||
-      build.pull_request.labels includes "omni-test"
+      build.pull_request.labels includes "nightly-test"
     steps:
       - label: ":full_moon: Omni · Perf Test · No Async Chunk"
+        source_file_dependencies: omni_qwen3_omni_perf
         key: nightly-omni-performance-no-async-chunk
         timeout_in_minutes: 180
         mirror_hardwares: a3_npu_4
@@ -24,6 +24,7 @@ steps:
             buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
             exit $$EXIT
       - label: ":full_moon: Omni · Perf Test · Async Chunk"
+        source_file_dependencies: omni_qwen3_omni_perf
         key: nightly-omni-performance-async-chunk
         timeout_in_minutes: 180
         mirror_hardwares: a3_npu_4
@@ -40,6 +41,7 @@ steps:
             buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
             exit $$EXIT
       - label: ":full_moon: Omni · MiniCPM-o 4.5 Test"
+        source_file_dependencies: omni_minicpmo_4_5_function
         mirror_hardwares: a3_npu_2
         env:
           VLLM_WORKER_MULTIPROC_METHOD: spawn
@@ -48,6 +50,7 @@ steps:
         commands:
           - pytest -s -v tests/e2e/offline_inference/test_minicpmo_4_5.py tests/e2e/online_serving/test_minicpmo_4_5.py tests/e2e/online_serving/test_minicpmo_4_5_expansion.py -m 'full_model' --run-level 'full_model'
       - label: ":full_moon: Omni · MiniCPM-o 4.5 · Duplex Test"
+        source_file_dependencies: omni_minicpmo_4_5_duplex_function
         timeout_in_minutes: 180
         mirror_hardwares: a3_npu_2
         env:
@@ -61,7 +64,13 @@ steps:
             tests/e2e/online_serving/test_minicpmo_4_5_duplex_expansion.py
             -m "npu and A3 and omni"
             --run-level full_model
+          - >-
+            pytest -s -v
+            tests/e2e/online_serving/test_minicpmo_4_5_window.py
+            -m "full_model and npu and A3 and omni"
+            --run-level full_model
       - label: ":full_moon: Omni · MiniCPM-o 4.5 · Accuracy Test"
+        source_file_dependencies: omni_minicpmo_4_5_accuracy
         timeout_in_minutes: 180
         mirror_hardwares: a3_npu_2
         env:
@@ -81,7 +90,27 @@ steps:
             EXIT=$$?
             buildkite-agent artifact upload "tests/e2e/accuracy/minicpmo_4_5/results/*.json"
             exit $$EXIT
+      - label: ":full_moon: Omni · Omni-DuplexEval CI Guard"
+        source_file_dependencies: omni_duplex_eval_ci
+        timeout_in_minutes: 240
+        mirror_hardwares: a3_npu_2
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+          HF_TOKEN: "${HF_TOKEN}"
+          VLLM_DUPLEX_EVAL_JUDGE_URL: "http://127.0.0.1:8001"
+        commands:
+          - |
+            set +e
+            pytest -s -v tests/e2e/accuracy/omni_duplex_eval/test_omni_duplex_eval_ci.py \
+              -m "full_model and npu and A3 and cards_2" \
+              --run-level full_model \
+              --basetemp /tmp/omni_duplex_ci
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/e2e/accuracy/omni_duplex_eval/results/*.json"
+            exit $$EXIT
       - label: ":full_moon: Omni · MiniCPM-o 4.5 · Perf Test"
+        source_file_dependencies: omni_minicpmo_4_5_perf
         key: nightly-omni-performance-minicpmo-4-5-npu
         timeout_in_minutes: 180
         mirror_hardwares: a3_npu_2
@@ -98,6 +127,7 @@ steps:
             buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
             exit $$EXIT
       - label: ":full_moon: Omni · MiniCPM-o 4.5 · Perf Test (A2)"
+        source_file_dependencies: omni_minicpmo_4_5_perf
         key: nightly-omni-performance-minicpmo-4-5-npu-a2
         timeout_in_minutes: 180
         mirror_hardwares: a2b3_npu_1
@@ -115,6 +145,7 @@ steps:
             buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
             exit $$EXIT
       - label: ":full_moon: Omni · MiniCPM-o 4.5 · Duplex Seed-TTS Perf Test"
+        source_file_dependencies: omni_minicpmo_4_5_duplex_perf
         key: nightly-omni-performance-minicpmo-4-5-duplex-seed-tts-npu
         timeout_in_minutes: 180
         mirror_hardwares: a3_npu_2
@@ -133,9 +164,10 @@ steps:
   - group: ":card_index_dividers: TTS Model Test"
     key: nightly-tts-test-group
     depends_on: upload-nightly-pipeline
-    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test" || build.pull_request.labels includes "tts-test"
+    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test"
     steps:
       - label: ":full_moon: TTS · Perf Test"
+        source_file_dependencies: tts_qwen3_tts_perf
         timeout_in_minutes: 180
         mirror_hardwares: a3_npu_2
         env:
@@ -151,15 +183,15 @@ steps:
             EXIT=$$?
             buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
             exit $$EXIT
-  - group: ":card_index_dividers: Diffusion I2I Model Test"
-    key: nightly-HunyuanImage3-x2i-test-group
+  - group: ":card_index_dividers: Diffusion X2I(&A&T) Model Test"
+    key: nightly-diffusion-x2iat-group
     depends_on: upload-nightly-pipeline
     if: >-
       build.env("NIGHTLY") == "1" ||
-      build.pull_request.labels includes "nightly-test" ||
-      build.pull_request.labels includes "diffusion-x2iat-test"
+      build.pull_request.labels includes "nightly-test"
     steps:
       - label: ":full_moon: HunyuanImage3-DIT · Accuracy Test"
+        source_file_dependencies: diffusion_hunyuan_image3_accuracy
         timeout_in_minutes: 180
         mirror_hardwares: a3_npu_4
         env:
@@ -169,15 +201,8 @@ steps:
           DIFFUSION_ATTENTION_BACKEND: TORCH_SDPA
         commands:
           - pytest -s -v tests/e2e/accuracy/test_hunyuan_image3_pixel_accuracy.py -m "full_model and A3" --run-level full_model
-  - group: ":card_index_dividers: Diffusion Benchmark Perf Test"
-    key: nightly-diffusion-benchmark-group
-    depends_on: upload-nightly-pipeline
-    if: >-
-      build.env("NIGHTLY") == "1" ||
-      build.pull_request.labels includes "nightly-test" ||
-      build.pull_request.labels includes "diffusion-x2iat-test"
-    steps:
       - label: ":full_moon: Diffusion Perf · HunyuanImage3 TP=2 CFGP=2"
+        source_file_dependencies: diffusion_hunyuan_image3_perf
         timeout_in_minutes: 180
         mirror_hardwares: a3_npu_4
         env:
@@ -194,6 +219,7 @@ steps:
             buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
             exit $$EXIT
       - label: ":full_moon: Diffusion Perf · HunyuanImage3 TP=2 SP=2"
+        source_file_dependencies: diffusion_hunyuan_image3_perf
         timeout_in_minutes: 180
         mirror_hardwares: a3_npu_4
         env:
@@ -210,6 +236,7 @@ steps:
             buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
             exit $$EXIT
       - label: ":full_moon: Diffusion Perf · HunyuanImage3 TP=4"
+        source_file_dependencies: diffusion_hunyuan_image3_perf
         timeout_in_minutes: 180
         mirror_hardwares: a3_npu_4
         env:
@@ -225,15 +252,81 @@ steps:
             buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
             buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
             exit $$EXIT
+      - label: ":full_moon: Diffusion X2I(&A&T) · Perf Test · Qwen-Image · Single-NPU"
+        source_file_dependencies: diffusion_qwen_image_perf
+        key: nightly-diffusion-x2iat-performance-qwen-image-single-npu
+        timeout_in_minutes: 180
+        mirror_hardwares: a3_npu_2
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          HF_TOKEN: "${HF_TOKEN}"
+          DIFFUSION_ATTENTION_BACKEND: TORCH_SDPA
+        commands:
+          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py -m "npu and A3 and cards_1" --test-config-file tests/dfx/perf/tests/test_qwen_image_vllm_omni.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
+            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
+            exit $$EXIT
+      - label: ":full_moon: Diffusion X2I(&A&T) · Perf Test · Qwen-Image · 4-NPU"
+        source_file_dependencies: diffusion_qwen_image_perf
+        key: nightly-diffusion-x2iat-performance-qwen-image-4npu
+        timeout_in_minutes: 180
+        mirror_hardwares: a3_npu_4
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          HF_TOKEN: "${HF_TOKEN}"
+          DIFFUSION_ATTENTION_BACKEND: TORCH_SDPA
+        commands:
+          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py -m "npu and A3 and cards_4" --test-config-file tests/dfx/perf/tests/test_qwen_image_vllm_omni.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
+            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
+            exit $$EXIT
   - group: ":card_index_dividers: Diffusion X2V Model Test"
     key: nightly-diffusion-x2v-group
     depends_on: upload-nightly-pipeline
     if: >-
       build.env("NIGHTLY") == "1" ||
-      build.pull_request.labels includes "nightly-test" ||
-      build.pull_request.labels includes "diffusion-x2v-test"
+      build.pull_request.labels includes "nightly-test"
     steps:
-      - label: ":full_moon: Diffusion X2V · Function Test"
+      - label: ":full_moon: Diffusion X2V · HunyuanVideo-1.5 Function Test"
+        source_file_dependencies: diffusion_hunyuan_video_function
+        timeout_in_minutes: 180
+        mirror_hardwares: a2b3_npu_4
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+          HF_TOKEN: "${HF_TOKEN}"
+          HF_HUB_OFFLINE: "1"
+        commands:
+          - pytest -s -v tests/e2e/online_serving/test_hunyuan_video_15_expansion.py -m "npu and A2" --run-level "full_model"
+      - label: ":full_moon: Diffusion X2V · HunyuanVideo-1.5 Perf Test"
+        source_file_dependencies: diffusion_hunyuan_video_perf
+        key: nightly-diffusion-x2v-performance-hunyuanvideo15
+        timeout_in_minutes: 180
+        mirror_hardwares: a3_npu_2
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+          HF_TOKEN: "${HF_TOKEN}"
+          HF_HUB_OFFLINE: "1"
+          DIFFUSION_ATTENTION_BACKEND: TORCH_SDPA
+        commands:
+          - export BENCHMARK_DIR=tests/dfx/perf/results
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py -m "npu and A3" --test-config-file tests/dfx/perf/tests/test_hunyuanvideo15_t2v_vllm_omni.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
+            exit $$EXIT
+      - label: ":full_moon: Diffusion X2V · Wan 2.2 Function Test"
+        source_file_dependencies: diffusion_wan22_function
         mirror_hardwares: a2b3_npu_4
         env:
           VLLM_WORKER_MULTIPROC_METHOD: spawn
@@ -241,7 +334,8 @@ steps:
           HF_TOKEN: "${HF_TOKEN}"
         commands:
           - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py -k npu_i2v --run-level "full_model"
-      - label: ":full_moon: Diffusion X2V · Perf Test"
+      - label: ":full_moon: Diffusion X2V · Wan 2.2 Perf Test"
+        source_file_dependencies: diffusion_wan22_perf
         key: nightly-diffusion-x2v-performance
         timeout_in_minutes: 180
         mirror_hardwares: a3_npu_4
@@ -251,11 +345,10 @@ steps:
           HF_TOKEN: "${HF_TOKEN}"
           DIFFUSION_ATTENTION_BACKEND: TORCH_SDPA
         commands:
-          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
+          - export BENCHMARK_DIR=tests/dfx/perf/results
           - |
             set +e
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py -m npu --test-config-file tests/dfx/perf/tests/test_wan22_i2v_vllm_omni.json
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py -m npu --test-config-file tests/dfx/perf/tests/test_wan22_i2v_vllm_omni.json
             EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
-            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
+            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
             exit $$EXIT

.buildkite/npu/test-npu-ready.yml

@@ -55,8 +55,8 @@ steps:
             -m 'core_model and npu and 310P'
             --run-level 'core_model'
 
-  - group: ":card_index_dividers: Omni Model Test"
-    key: ready-omni-group
+  - group: ":card_index_dividers: E2E Test"
+    key: ready-e2e-group
     depends_on: upload-ready-pipeline
     steps:
       - label: "Omni · MiniCPM-o 4.5 Test"
@@ -66,6 +66,7 @@ steps:
           VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
         commands:
           - pytest -s -v tests/e2e/offline_inference/test_minicpmo_4_5.py tests/e2e/online_serving/test_minicpmo_4_5.py -m 'core_model and npu and A3 and omni' --run-level 'core_model'
+
       - label: "Omni · MiniCPM-o 4.5 Duplex Test"
         mirror_hardwares: a3_npu_2
         env:
@@ -73,6 +74,7 @@ steps:
           VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
         commands:
           - pytest -s -v tests/e2e/online_serving/test_minicpmo_4_5_duplex.py -m 'core_model and npu and A3 and omni' --run-level 'core_model'
+
       - label: "Omni · MiniCPM-o 4.5 Perf Test"
         timeout_in_minutes: 60
         mirror_hardwares: a3_npu_2
@@ -83,10 +85,15 @@ steps:
           - export BENCHMARK_DIR=tests/dfx/perf/results
           - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_minicpmo_4_5_ready.json -m 'core_model and npu and A3 and omni'
 
-  - group: ":card_index_dividers: TTS Model Test"
-    key: ready-tts-group
-    depends_on: upload-ready-pipeline
-    steps:
+      - label: "Omni · Qwen3-Omni Test"
+        mirror_hardwares: a3_npu_4
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+        commands:
+          - pytest -s -v tests/e2e/online_serving/test_qwen3_omni.py -m 'core_model and npu and A3 and omni' --run-level 'core_model'
+        timeout_in_minutes: 60
+
       - label: ":sound: TTS · Qwen3-TTS CustomVoice Test"
         mirror_hardwares: a3_npu_2
         env:
@@ -95,3 +102,55 @@ steps:
           HF_TOKEN: "${HF_TOKEN}"
         commands:
           - pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py -m 'core_model and npu' --run-level 'core_model'
+
+      - label: "Diffusion · HunyuanVideo-1.5 Test"
+        mirror_hardwares: a3_npu_2
+        env:
+            VLLM_WORKER_MULTIPROC_METHOD: spawn
+            VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+        commands:
+            - |
+              timeout 40m bash -c "
+                export VLLM_IMAGE_FETCH_TIMEOUT=60
+                pytest -s -v tests/e2e/online_serving/test_hunyuan_video_15_expansion.py -m 'core_model and diffusion and npu' --run-level 'core_model'
+                "
+        timeout_in_minutes: 60
+
+      - label: "Diffusion · Qwen Image Test"
+        mirror_hardwares: a3_npu_2
+        env:
+            VLLM_WORKER_MULTIPROC_METHOD: spawn
+            VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+        commands:
+            - |
+                timeout 40m bash -c "
+                export VLLM_IMAGE_FETCH_TIMEOUT=60
+                pytest -s -v tests/e2e/online_serving/test_qwen_image.py -m 'core_model and npu' --run-level 'core_model'
+                "
+        timeout_in_minutes: 60
+
+      - label: "Diffusion · Wan22 Test"
+        mirror_hardwares: a3_npu_2
+        env:
+            VLLM_WORKER_MULTIPROC_METHOD: spawn
+            VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+        commands:
+            - |
+                timeout 40m bash -c "
+                export VLLM_IMAGE_FETCH_TIMEOUT=60
+                pytest -s -v tests/e2e/online_serving/test_wan22_t2v.py -m 'core_model and npu' --run-level 'core_model'
+                "
+        timeout_in_minutes: 60
+
+      - label: "Diffusion · Hunyuan Image Test"
+        mirror_hardwares: a3_npu_4
+        env:
+            VLLM_WORKER_MULTIPROC_METHOD: spawn
+            VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+        commands:
+            - |
+                timeout 40m bash -c "
+                export VLLM_IMAGE_FETCH_TIMEOUT=60
+                pytest -s -v tests/e2e/online_serving/test_hunyuan_image.py -m 'core_model and diffusion and npu' --run-level 'core_model'
+                "
+        timeout_in_minutes: 60

.claude/skills/3d-animation-short-generator

@@ -0,0 +1 @@
+../../.agents/skills/3d-animation-short-generator

.claude/skills/brand-promo-video-generator

@@ -0,0 +1 @@
+../../.agents/skills/brand-promo-video-generator

.claude/skills/co-op-game-intro-generator

@@ -0,0 +1 @@
+../../.agents/skills/co-op-game-intro-generator

.claude/skills/h3-prompt-writing

@@ -0,0 +1 @@
+../../.agents/skills/h3-prompt-writing

.claude/skills/handdrawn-live-video-generator

@@ -0,0 +1 @@
+../../.agents/skills/handdrawn-live-video-generator

.claude/skills/minimalist-product-ad-generator

@@ -0,0 +1 @@
+../../.agents/skills/minimalist-product-ad-generator

.claude/skills/music-video-subtitle-generator

@@ -0,0 +1 @@
+../../.agents/skills/music-video-subtitle-generator

.claude/skills/paper-collage-explainer-generator

@@ -0,0 +1 @@
+../../.agents/skills/paper-collage-explainer-generator

.claude/skills/papercraft-stop-motion-explainer

@@ -0,0 +1 @@
+../../.agents/skills/papercraft-stop-motion-explainer

.claude/skills/production-add-diffusion-model/references/production-validation.md

@@ -340,7 +340,7 @@ On supported GPU hosts, wrap a test or soak with the repository monitor:
 
 ```bash
 bash tests/dfx/stability/scripts/resource_monitor.sh run --backend gpu -- \
-  pytest -s -v tests/dfx/stability/scripts/test_stability_<model>.py -m slow
+  pytest -s -v tests/dfx/stability/scripts/run_stability_<model>.py -m slow
 ```
 
 Check the target revision's monitor backends. Do not claim CPU/NPU monitoring

.claude/skills/readme.md

@@ -60,6 +60,14 @@ self-check.
   `test-weekly.yml`), and copy-paste local plus CI-like `pytest` commands; see
   `references/test-routing.md` for level-to-command mapping
 
+## MiniMax H3 Creative Skills
+
+The [portable H3 collection](../../.agents/skills/README.md) contains the prompt
+writing skill and eight creative workflows for product ads, 3D shorts, paper art,
+brand promos, music videos, co-op intros, and hand-drawn/live-action fusion.
+Same-named directories here link to the canonical `.agents/skills/` copies, so
+Claude Code and Codex use the same instructions and templates.
+
 ## Maintenance Guidelines
 
 - Keep skill names short and task-oriented.

.claude/skills/vllm-omni-test/SKILL.md

@@ -468,10 +468,10 @@ When the user asks for **invalid parameter validation**, **invalid request bodie
 | **Invalid parameters Test · H100** | `pytest -s -v tests/dfx/reliability/invalid_param_test/ -m "slow and H100"` | Diffusion / omni / video invalid-param tests with `H100` hardware mark |
 | **Invalid parameters Test · L4** | `pytest -s -v tests/dfx/reliability/invalid_param_test/ -m "slow and L4"` | TTS / lighter models marked `L4` |
 
-- **Trigger:** `build.env("WEEKLY") == "1"` or PR label `weekly-test`.
+- **Trigger:** `build.env("WEEKLY") == "1"` (all weekly jobs) or PR label `weekly-test` (Reliability / Perf, then filtered by `source_file_dependencies`).
 - **Default:** extending an existing `invalid_param_test` script needs **no YAML edit** — the weekly steps already sweep the whole directory.
 - **Edit YAML only** when adding a **new hardware queue**, a **new top-level script** that must run in isolation, or a **model-specific weekly shard** (mirror neighboring reliability steps).
-- Weekly steps **do not** use `source_file_dependencies`.
+- Weekly Reliability / Perf jobs declare `source_file_dependencies`. Filtering applies only on the `weekly-test` label; `WEEKLY=1` on `main` keeps every job in those groups. CPU coverage (`WEEKLY=1` only) and the E2E Tests group (`NON_CRITICAL=1` only) do **not** use this field.
 
 **Example — append to `test_invalid_image_generation.py`:**
 
@@ -644,7 +644,7 @@ If the test is not already collected by an existing pipeline command (for exampl
 | **L2** | `test-ready.yml` with `core_model` + `source_file_dependencies` | merge / nightly pipelines |
 | **Invalid param** | `test-weekly.yml` — **Invalid parameters Test** group (`-m "slow and H100"` / `-m "slow and L4"`). Usually **no YAML edit** when appending cases to existing `invalid_param_test` scripts | `test-ready.yml`, `test-merge.yml`, `test-nightly.yml` |
 
-`test-nightly.yml` shards use **explicit pytest file paths** in `commands` (and PR labels like `diffusion-x2iat-test`); they generally **do not** use `source_file_dependencies`. Only suggest merge/ready YAML when the requested level is L3/L2 (or the user explicitly asks for multi-level CI).
+`test-nightly.yml` shards use **explicit pytest file paths** in `commands` plus `source_file_dependencies`; PR trigger is `nightly-test` (path filter), or `NIGHTLY=1` on `main` (all jobs). Only suggest merge/ready YAML when the requested level is L3/L2 (or the user explicitly asks for multi-level CI).
 
 **`test-nightly.yml` top-level groups** (each uses `full_model` + type marker in pytest `-m`):
 
@@ -653,7 +653,7 @@ If the test is not already collected by an existing pipeline command (for exampl
 | **Omni Model Test** | `omni` | Function / doc / accuracy / **perf** for Qwen-Omni family |
 | **TTS Model Test** | `tts` | `-m "full_model and L4 and tts"` (and H100 variants when enabled); **Perf Test** uses `run_benchmark.py` + `test_tts.json` |
 | **Diffusion X2I(&A&T) Model Test** | `diffusion` | Image/audio/text diffusion; explicit file shards for **Function**; **Perf Test · <Model>** uses `run_diffusion_benchmark.py` + JSON config |
-| **Diffusion X2V Model Test** | `diffusion` | Video-only (Wan, HunyuanVideo, …); PR label `diffusion-x2v-test`; separate **Perf** steps when in-tree |
+| **Diffusion X2V Model Test** | `diffusion` | Video-only (Wan, HunyuanVideo, …); separate **Perf** steps when in-tree |
 
 Guidelines when editing YAML:
 

.claude/skills/vllm-omni-test/references/test-routing.md

@@ -37,7 +37,7 @@ Every model-centric e2e test must declare **exactly one** type marker:
 | **Diffusion X2I(&A&T) Model Test** | x2**i** / x2**a** / x2**t** — image, audio, text (non-video) | Qwen-Image*, BAGEL, FLUX, SD3, Z-Image, LongCat, DreamZero |
 | **Diffusion X2V Model Test** | x2**v** — video only | Wan2.2, HunyuanVideo 1.5, Wan VACE |
 
-PR labels for selective nightly runs: `diffusion-x2iat-test`, `diffusion-x2v-test` (plus `nightly-test` / `NIGHTLY=1`).
+PR label for L4 on a branch: `nightly-test` (jobs then filtered by `source_file_dependencies`). Scheduled `main` uses `NIGHTLY=1` and runs every nightly job. L5 Reliability / Perf use `weekly-test` with the same path filter; CPU coverage and the weekly E2E group are env-only (`WEEKLY=1` / `NON_CRITICAL=1`) and are not source-filtered.
 
 ## Level and Marker Mapping
 
@@ -238,7 +238,7 @@ pytest -s -v dfx/reliability/invalid_param_test/ -m "slow and L4"
 pytest -s -v dfx/reliability/invalid_param_test/test_invalid_image_generation.py::test_images_generations_invalid_requests -m "slow and H100"
 ```
 
-**Trigger:** `WEEKLY=1` or PR label `weekly-test`. Appending cases to existing scripts usually needs **no YAML edit** (directory sweep). No `source_file_dependencies` on weekly steps.
+**Trigger:** `WEEKLY=1` (all jobs) or PR label `weekly-test` (Reliability / Perf, filtered by `source_file_dependencies`). Appending cases to existing scripts usually needs **no YAML edit** (directory sweep).
 
 ### Platform-targeted examples
 

.github/CODEOWNERS

@@ -67,6 +67,11 @@
 # File-level overrides within entrypoints/
 /vllm_omni/entrypoints/openai/errors.py @alex-jw-brooks @NickCao
 /vllm_omni/entrypoints/cli/benchmark/ @alex-jw-brooks @Bounty-hunter
+# Shared wire codec: protocol/realtime (OpenAI Realtime GA) and protocol/duplex
+# (vLLM-Omni's extension of it) — RFC #6592 P0a. Layer-neutral by design — the
+# engine depends on it — so it is owned by the serving-boundary document rather
+# than by any one consumer. — docs/design/module/entrypoints.md
+/vllm_omni/protocol/ @alex-jw-brooks @linyueqian @NickCao
 
 # Error contracts — docs/design/module/error_contracts.md
 /vllm_omni/errors.py @alex-jw-brooks @NickCao

.github/PULL_REQUEST_TEMPLATE.md

@@ -12,4 +12,7 @@ PLEASE FILL IN THE PR DESCRIPTION HERE.
 ## Test Result
 
 **BEFORE SUBMITTING:** read [CONTRIBUTING.md](https://github.com/vllm-project/vllm-omni/blob/main/CONTRIBUTING.md) and run the [precheck-pr skill](https://github.com/vllm-project/vllm-omni/blob/main/.claude/skills/precheck-pr/SKILL.md) with the code agent for a self-check against project conventions.
+
+> [!NOTE]
+> To avoid agent spam, vLLM-Omni limits contributors without write access to **5 open PRs**. If this blocks well-intentioned critical work, please contact a committer to request bypass list access.
 (anything written below this line will be removed by GitHub Actions)

.github/workflows/build_wheel.yml

@@ -56,6 +56,35 @@ jobs:
         run: |
           bash scripts/build_wheel.sh --python python
 
+      - name: Validate runtime package data from source archive
+        run: |
+          set -euo pipefail
+          archive_dir="$(mktemp -d)"
+          install_dir="$(mktemp -d)"
+          vcs_wheels=(dist/*.whl)
+          if [[ ${#vcs_wheels[@]} -ne 1 || ! -f "${vcs_wheels[0]}" ]]; then
+            echo "Expected exactly one VCS-built wheel" >&2
+            exit 1
+          fi
+          git archive HEAD | tar -x -C "${archive_dir}"
+          (
+            cd "${archive_dir}"
+            VLLM_OMNI_TARGET_DEVICE=cuda \
+              VLLM_OMNI_VERSION_OVERRIDE=0.0.0 \
+              python -m build --wheel --outdir archive-dist
+          )
+          wheels=("${archive_dir}"/archive-dist/*.whl)
+          if [[ ${#wheels[@]} -ne 1 || ! -f "${wheels[0]}" ]]; then
+            echo "Expected exactly one archive-built wheel" >&2
+            exit 1
+          fi
+          python -m pip install --no-deps --target "${install_dir}" "${wheels[0]}"
+          python tools/check_wheel_package_data.py \
+            --source-root "${archive_dir}" \
+            --wheel "${wheels[0]}" \
+            --vcs-wheel "${vcs_wheels[0]}" \
+            --installed-root "${install_dir}"
+
       - name: Upload Wheel Artifact
         uses: actions/upload-artifact@v7
         with:

.gitignore

@@ -243,9 +243,15 @@ datasets/
 *.csv
 *.json
 !recipes/cosmos3/negative_prompt.json
+!benchmarks/mammoth_moda2/reference_results.json
 !tests/dfx/perf/tests/*.json
 !apps/ComfyUI-vLLM-Omni/example_workflows/*.json
+# Omni-DuplexEval CI config JSON (e.g. accuracy/omni_duplex_eval)
+!tests/e2e/accuracy/omni_duplex_eval/*.json
 *.jsonl
+# Shared Hunyuan IT2I input for accuracy tests and vllm bench.
+!tests/assets/hunyuan_image3/it2i.jsonl
+!tests/assets/livekit/client_vad_replay.jsonl
 *.parquet
 
 # Output files

CONTRIBUTING.md

@@ -11,3 +11,5 @@ how to extend allowlists) are documented in the
 section of that guide.
 
 Before submitting a PR, run the [precheck-pr skill](https://github.com/vllm-project/vllm-omni/blob/main/.claude/skills/precheck-pr/SKILL.md) with the code agent for a self-check against project conventions.
+
+vLLM-Omni limits contributors without write access to **5 open PRs**; see the [Pull Request Limit](docs/contributing/README.md#pull-request-limit) section of the contributing guide.

README.md

@@ -16,6 +16,7 @@ Easy, fast, and cheap omni-modality model serving for everyone
 
 *Latest News* 🔥
 
+- [2026/09] We released [0.30.0](https://github.com/vllm-project/vllm-omni/releases/tag/v0.30.0), rebased onto vLLM 0.30.0, featuring a unified full-duplex serving framework around engine-owned sessions for [MiniCPM-o 4.5](recipes/OpenBMB/MiniCPM-o-4_5.md) and AURA, native cross-stage KV and multimodal payload transfer (Mooncake AR-to-DiT handoff and NIXL connectors), interactive world-model serving with [LingBot World](recipes/Robbyant/LingBot-World-2.0.md), and realtime [MiniMax H3](recipes/MiniMaxAI/MiniMax-H3.md) Turbo inference on NVIDIA Blackwell and Ascend 950.
 - [2026/08] We released [0.28.0](https://github.com/vllm-project/vllm-omni/releases/tag/v0.28.0), featuring production-ready [MiniMax H3](recipes/MiniMaxAI/MiniMax-H3.md) serving on GPU and NPU, a unified AR/DiT paged KV cache runtime, and enhanced realtime full-duplex serving for the [MiniCPM-o series](recipes/OpenBMB/MiniCPM-o-4_5.md).
 - [2026/08] [VeRL-Omni](https://github.com/verl-project/verl-omni) `v0.2.0` is released: faster diffusion RL powered by vLLM-Omni (request-level/step-wise batching with FA3), rebuilt Qwen3-Omni multimodal training (DPO & GSPO), plus LTX-2.3, Qwen-Image-Edit support and more. See the [release notes](https://github.com/verl-project/verl-omni/releases/tag/v0.2.0).
 - [2026/08] We released [0.26.0](https://github.com/vllm-project/vllm-omni/releases/tag/v0.26.0) - aligned with the vLLM 0.26 release line, featuring [MiniMax H3](recipes/MiniMaxAI/MiniMax-H3.md) joint video/audio generation, an experimental full-duplex realtime runtime for [MiniCPM-o 4.5](recipes/OpenBMB/MiniCPM-o-4_5.md), distributed layerwise diffusion offload, and broader model, hardware, streaming, TTS, and quantization support.
@@ -58,9 +59,9 @@ vLLM-Omni is flexible and easy to use with:
 vLLM-Omni seamlessly supports most popular open-source models on HuggingFace, including:
 
 - **Omni-modality models** (e.g. Qwen3-Omni, MiniCPM-o 4.5, Cosmos3, HunyuanImage, BAGEL)
-- **TTS models** (e.g. Qwen3-TTS, IndexTTS 2.5, dots.tts, CosyVoice3)
-- **Diffusion models** — image, video, and audio generation (e.g. MiniMax H3, LTX-2.5, SANA-Video, Wan2.2)
-- **Robot-policy and action models** (e.g. π0, GR00T-N1.7, DreamZero-DROID, InternVLA-A1)
+- **TTS models** (e.g. Qwen3-TTS, Tencent AuK, Breeze-TTS-2, CosyVoice3)
+- **Diffusion models** — image, video, and audio generation (e.g. MiniMax H3, LingBot World, MAGI-2, LTX-2.5, Wan2.2)
+- **Robot-policy and action models** (e.g. π0.5, GR00T-N1.7, DreamZero-DROID, InternVLA-A1)
 
 ## Getting Started
 

apps/ComfyUI-vLLM-Omni/README.md

@@ -13,7 +13,6 @@ It can send model inference requests to either a locally running vLLM-Omni servi
 > [!TIP]
 > If you run both ComfyUI and vLLM-Omni on the same device, you can create separate virtual environments and use different Python versions for them.
 
-
 ## Installation
 
 Copy this folder to the `custom_nodes` subfolder of your ComfyUI installation. Your directory should look like `ComfyUI/custom_nodes/ComfyUI-vLLM-Omni`.
@@ -21,9 +20,10 @@ Copy this folder to the `custom_nodes` subfolder of your ComfyUI installation. Y
 If you are running ComfyUI during copying, you should restart ComfyUI to load this extension.
 
 > [!TIP]
-> You can use utility websites such as https://download-directory.github.io/ to download a subdirectory of a repo. Also checkout community discussions (e.g., https://stackoverflow.com/questions/7106012/download-a-single-folder-or-directory-from-a-github-repository) for more info.
+> You can use utility websites such as <https://download-directory.github.io/> to download a subdirectory of a repo. Also checkout community discussions (e.g., <https://stackoverflow.com/questions/7106012/download-a-single-folder-or-directory-from-a-github-repository>) for more info.
 
 On the device and virtual environment you run ComfyUI, launch ComfyUI with
+
 ```bash
 cd ComfyUI
 
@@ -35,6 +35,7 @@ python main.py --cpu
 ```
 
 On the device and virtual environment you run vLLM-Omni, start a model service with
+
 ```bash
 vllm serve The_Model_ID_to_Serve --omni --port 8000
 ```
@@ -48,14 +49,31 @@ This extension offers the following nodes based on the output modalities (at **C
 
 - **Generate Image** for text-to-image and image-to-image tasks
 - **Generate Video** for text-to-video, first-frame/image-to-video, and reference-conditioned video
+- **Latent Mask Editing** for MiniMax-H3 latent-mask editing (source media plus video/audio noise masks)
+- **FastH3 Deployment** for routing text-to-video requests to a MiniMax-H3 server with FastH3 fused at startup
 - **Multimodality Understanding** for multimodality-to-text and multimodality-to-audio tasks
 - **TTS** and **TTS Voice Clone** for TTS tasks
+- **Generate Music** for lyrics-and-description-to-music generation
 
 This extension also offers example workflows (at **ComfyUI sidebar -> Templates -> vLLM-Omni**)
 
 > [!NOTE]
 > The node UI and feature designs are intended to match vLLM-Omni online serving interfaces. It cannot offer more than what the interfaces support.
 
+Every node carries the vLLM-Omni mark in its title bar and is tinted by what it outputs, so a graph is readable at a glance:
+
+| Colour | Nodes | What they produce |
+| --- | --- | --- |
+| Blue | Generate Image, Generate Video, Multimodality Understanding, TTS, TTS Voice Clone | A generated image, video, audio, or text. These are the only nodes that reach a server. |
+| Amber | AR / Diffusion / Multi-Stage Sampling Params | Sampling parameters that apply to any model |
+| Purple | Qwen TTS Params, Wan Video Params, MiniMax-H3 Video Params | Parameters that only one model family accepts |
+| Red | LoRA, FastH3 Deployment | Which weights the server is expected to have loaded |
+| Teal | Video References, Latent Mask Editing | Reference media |
+
+Recolouring a node by hand (right click -> Colors) overrides its tint, and the choice is kept.
+
+**Generate Video** takes a clip length in seconds (`duration`), not a frame count. Frames stay the wire unit and are derived with the node's `fps`, so the length is always measured against the rate that is actually served; models that accept only certain frame counts still round to their own lattice server-side. Graphs saved before this widget existed stored `num_frames` in its place and are converted on load, using the fps recorded alongside it -- the browser console names every node it rewrites.
+
 To build a simple workflow yourself,
 
 - Drag a generation node onto the canvas.
@@ -117,11 +135,146 @@ You can configure per-stage sampling parameters for multi-stage models.
 </p>
 
 > [!TIP]
-> Connect a **frame** image for first-frame / image-to-video (e.g., Wan I2V, MiniMax-H3 FL2VA).
+> Connect **frame** for backward-compatible first-frame / image-to-video behavior. For MiniMax-H3 FL2VA,
+> connect **first_frame**, **last_frame**, or both to condition the start frame, end frame, or both.
 >
 > For reference-conditioned generation (MiniMax-H3 Ref2VA), connect a **Video References** node instead.
 >
-> Do not use `frame` and `references` together. Task routing is automatic from which inputs you connect.
+> Do not combine `frame` with `first_frame` or `last_frame`, and do not combine any frame input with
+> `references`. Task routing is automatic from which inputs you connect.
+
+For MiniMax-H3 Ref2VA, **Video References** accepts up to 9 images (`image_1`–`image_9`),
+3 videos (`video_1`–`video_3`), and 3 audio clips (`audio_1`–`audio_3`), with at most
+12 connected inputs in total. Any mixture containing at least one image or video
+is supported; empty and audio-only references are rejected. For example, you can
+combine 6 images, 3 videos, and 3 audio clips in one request.
+
+Within each media type, references follow slot-number order, skipping unconnected
+slots. For example, connecting `image_2` and `image_9` sends `image_2` as the first
+image and `image_9` as the second. Each image slot uses the first image in its batch.
+Existing connections to `image_1`, `image_2`, `audio_1`, `audio_2`, `video_1`, and
+`video_2` remain valid in saved workflows.
+
+#### MiniMax-H3 Reference to Video
+
+Load [MiniMax-H3 Reference to Video](example_workflows/vLLM-Omni%20MiniMax-H3%20Reference%20to%20Video.json)
+from **Templates → ComfyUI-vLLM-Omni**, or drag the JSON onto the canvas.
+It adapts the [official ComfyUI R2V workflow](https://github.com/Comfy-Org/workflow_templates/blob/main/templates/video_minimax_h3_r2v.json)
+to remote vLLM-Omni execution. Model loading, sampling, and VAE decoding run on
+the server; ComfyUI loads references and saves the returned video.
+
+Use an extension version containing the audio-preserving output fix from
+[#7456](https://github.com/vllm-project/vllm-omni/pull/7456). Older versions can
+drop the server's audio while decoding the response.
+
+1. Start a Ref2VA-capable service using the [MiniMax-H3 recipe](../../recipes/MiniMaxAI/MiniMax-H3.md).
+   Set **Generate Video** to its `/v1` URL and served model name.
+2. Select your own image in **Load Image**. The template's filenames are placeholders
+   relative to ComfyUI's input directory; no sample assets or model weights are bundled.
+3. Connect **Load Video** and **Load Audio** when needed. Duplicate the loaders to
+   fill more reference slots, or disconnect the image when using video-only input.
+   Keep the references connected to **Generate Video** and leave `frame` disconnected.
+4. Match prompt tags such as `<Picture 1>`, `<Video 1>`, and `<Audio 1>` to the
+   connected references, counting each media type separately and skipping empty slots.
+5. Run the workflow. **Save Video** writes an MP4 under `output/video/` and preserves
+   the generated audio when the audio-output prerequisite is installed.
+
+The default is **1344×768, 24 FPS, 124 frames** (about 5.17 seconds), 50 sampling
+points, video flow shift 12, audio flow shift 3, and seed 42. The duration widget
+is set to 5.167 seconds, which converts to 124 frames at 24 FPS. Other H3 canvas presets
+are 1024×768, 768×768, 768×1024, and 768×1344. Keep frame counts at `17k+5` within
+the 4–15 second output range; examples are 107, 124, 209, and 345 frames.
+Reference video/audio clips must each be 2–15 seconds, with at most 15 seconds of
+reference video and at most 15 seconds of standalone reference audio per request.
+
+For **Ref2VA Turbo**, use the same template. Configure a Ref2VA-only server with
+`--lora-backend peft --lora-path "$TURBO_LORA" --task-type ref2va`, following the
+recipe's hardware setup. For `minimax_h3_ref2v_turbo_4step_v0.1_bf16.safetensors`,
+set **LoRA** `local_path` to that file's path on the server and connect it to
+**Generate Video**. Use `num_inference_steps=5`, `flow_shift=12`,
+`audio_flow_shift=3`, and LoRA `scale=1`. The API counts sigma points, so five
+points produce four denoiser evaluations. Use the Diffusers artifact, not its
+`_comfyui_` export. To return to base mode, disconnect LoRA and restore 50 steps.
+The 8-step v1.0 Ref2VA adapter instead requires nine sampling points and video
+flow shift 6. FastH3 is a separate startup-fused T2VA deployment.
+
+Validate the template's wiring and defaults locally with:
+
+```bash
+python -m pytest tests/e2e/features/comfyui/test_h3_reference_workflow.py -q
+```
+
+For a real-model check, run the workflow against H3 and inspect the saved file:
+
+```bash
+ffprobe -v error -show_entries stream=codec_type,width,height,r_frame_rate,sample_rate,channels \
+  -show_entries format=duration -of json "$OUTPUT_MP4"
+```
+
+Check for 24 FPS video, stereo audio, the requested canvas, and aligned duration;
+also play the result to assess reference conditioning and audio/video synchronization.
+Record the server commit, model/adapter, input assets, prompt, settings, and output
+alongside the result. Local schema or mocked-server checks do not establish H3
+generation quality or replace this real-model validation.
+
+### MiniMax H3 Text to Video
+
+Import [the H3 text-to-video template](example_workflows/MiniMax_H3_Text_to_Video.json)
+for remote video generation with audio. It includes connected base presets and
+optional Turbo sampling, H3 parameters, and Remote LoRA nodes. See the
+[workflow guide](docs/minimax-h3-t2v.md) for server setup, artifact-specific Turbo
+settings, and saved-video/audio validation. Set Remote LoRA’s `local_path` to
+the downloaded Turbo artifact on your server before enabling Turbo.
+
+#### H3 video upscale (WF-07)
+
+The **vLLM-Omni MiniMax H3 Video Upscale** template generates video remotely, upscales it with SeedVR2, and saves the original and upscaled videos with the generated audio and FPS. See [workflow setup](docs/wf07-h3-upscale.md).
+
+#### FastH3 text-to-video
+
+FastH3 is fused into MiniMax-H3 when the vLLM-Omni server starts; it is not a request-switchable LoRA. Download one adapter variant and start a non-offloaded FL2VA server. For the Dense / Data-Free profile:
+
+```bash
+export FASTH3_DIR=/path/to/fasth3
+hf download FastVideo/FastVideo-FastH3-4-step-Preview-v1-LoRA \
+  dense-datafree/adapter_model.safetensors --local-dir "${FASTH3_DIR}"
+
+vllm serve /path/to/MiniMax-H3 \
+  --omni \
+  --task-type fl2va \
+  --lora-path "${FASTH3_DIR}/dense-datafree/adapter_model.safetensors" \
+  --port 8000
+```
+
+For the VSA / Data-Free profile, download `vsa-datafree/adapter_model.safetensors`, use that file as `--lora-path`, and add:
+
+```bash
+--diffusion-attention-backend FASTVIDEO_VSA \
+--fastvideo-vsa-topk 64
+```
+
+The VSA profile also requires a compatible `fastvideo-kernel` installation. See the [MiniMax-H3 FastH3 recipe](../../recipes/MiniMaxAI/MiniMax-H3.md#fasth3-adapter) for the full serving contract, supported parallel layouts, and kernel requirements.
+
+Open the **vLLM-Omni FastH3 Text to Video** template, then:
+
+- Set the server URL and served model name on **FastH3 Deployment**.
+- Connect its output to **Generate Video → fast_h3**. When connected, the deployment node's URL and model take precedence over the corresponding Generate Video widgets.
+- Keep `frame`, `references`, **LoRA**, and **MiniMax-H3 Video Params** disconnected. FastH3 Preview v1 supports T2VA only, is already fused, and owns both flow shifts.
+- A connected **Diffusion Sampling Params** node may set seed and other ordinary sampling options. The integration always enforces four denoising steps and 24 FPS for FastH3.
+
+The node records which server the workflow targets; it does not start one, nor switch adapters or attention backends on a running server.
+
+#### Latent-mask editing (MiniMax-H3)
+
+The [WF-05 template](example_workflows/vLLM-Omni%20MiniMax-H3%20Latent%20Mask%20Editing.json) contains inpainting, object removal, continuation, and extension examples in one graph. See the [workflow guide](docs/wf05-h3-latent-editing.md) for inputs, mask settings, dependencies, and preview limitations.
+
+Connect a **Latent Mask Editing** node to **Generate Video → latent_edit** to edit a source clip instead of generating from scratch. It uploads the source media and serializes the video/audio noise masks the MiniMax H3 API accepts:
+
+- `source_video` / `source_audio` — the media to edit.
+- `video_mask` — a ComfyUI mask image; `0` preserves a region, `1` regenerates it, fractional values blend. A 2D mask `[H, W]` is applied to every frame; a 3D mask `[T, H, W]` is treated as a temporal mask (one slice per frame) for continuation or extension. The client area-downsamples it by 16 to keep the upload small, and the server resizes it to the video latent grid.
+- `audio_mask` — a scalar in `[0, 1]`; `0` keeps the source audio, `1` regenerates it, fractional values blend.
+
+The node requires at least one mask; the server then enforces the cross-field rules (a source requires a matching mask, and a non-trivial mask requires its source). This node only forwards inputs to the server; the served model must declare latent-mask editing support (MiniMax-H3).
 
 ### TTS (e.g., Qwen TTS series)
 
@@ -137,6 +290,17 @@ You can configure per-stage sampling parameters for multi-stage models.
 > [!TIP]
 > There is a dedicated node for VoiceClone tasks with reference audio input. Other simple text-to-speech tasks should use the regular TTS node.
 
+### Music generation (e.g., MiniMax Music 3)
+
+(Also available at **ComfyUI sidebar->Template->vLLM-Omni->vLLM-Omni Music Generation**)
+
+<p align="center">
+  <picture>
+    <source media="(prefers-color-scheme: dark)" src="https://raw.githubusercontent.com/vllm-project/vllm-omni/refs/heads/main/apps/ComfyUI-vLLM-Omni/docs/images/comfyui-music-generation.jpg">
+    <img alt="vLLM-Omni Music Generation" src="https://raw.githubusercontent.com/vllm-project/vllm-omni/refs/heads/main/apps/ComfyUI-vLLM-Omni/docs/images/comfyui-music-generation.jpg" width=55%>
+  </picture>
+</p>
+
 ### Chaining multiple model services
 
 (Also available at **ComfyUI sidebar->Template->vLLM-Omni->vLLM-Omni Chaining Services**)
@@ -152,6 +316,8 @@ You can configure per-stage sampling parameters for multi-stage models.
 
 Follow the [development convention and rules of vLLM-Omni](https://docs.vllm.ai/projects/vllm-omni/en/latest/contributing/).
 
+Node tints and the title-bar mark are applied in `web/main.js`, keyed off each node's declared output types rather than a list of node names. A new node that returns an existing type is themed with no front-end change; a new output type needs one entry in `FAMILY_BY_OUTPUT` there.
+
 ## Limitation and Non-Goals
 
 - Single server mode only. No automatic load balancing or failover.
@@ -178,11 +344,11 @@ Whenever you find an issue or problem, please
 
 Features
 
-- https://github.com/dougbtv/comfyui-vllm-omni/ The official reference implementation for ComfyUI integration with vLLM-Omni's DALL-E compatible image generation API.
-- https://github.com/Comfy-Org/ComfyUI/tree/master/comfy_extras ComfyUI's built-in node implementations.
+- <https://github.com/dougbtv/comfyui-vllm-omni/> The official reference implementation for ComfyUI integration with vLLM-Omni's DALL-E compatible image generation API.
+- <https://github.com/Comfy-Org/ComfyUI/tree/master/comfy_extras> ComfyUI's built-in node implementations.
 
 UI/UX design references
 
-- https://github.com/sgl-project/sglang/pull/15271 SGLang Diffusion's official ComfyUI integration for image and video generation.
-- https://github.com/SXQBW/ComfyUI-Qwen-Omni A third party ComfyUI integration for Qwen Omni series.
-- https://github.com/flybirdxx/ComfyUI-Qwen-TTS https://github.com/DarioFT/ComfyUI-Qwen3-TTS Third  party ComfyUI integrations for Qwen TTS series.
+- <https://github.com/sgl-project/sglang/pull/15271> SGLang Diffusion's official ComfyUI integration for image and video generation.
+- <https://github.com/SXQBW/ComfyUI-Qwen-Omni> A third party ComfyUI integration for Qwen Omni series.
+- <https://github.com/flybirdxx/ComfyUI-Qwen-TTS> <https://github.com/DarioFT/ComfyUI-Qwen3-TTS> Third  party ComfyUI integrations for Qwen TTS series.

apps/ComfyUI-vLLM-Omni/__init__.py

@@ -1,4 +1,8 @@
-"""Top-level package for comfyui_vllm_omni."""  # noqa: N999  # This is not a python library intended to be imported
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+# ruff: noqa: N999
+
+"""Top-level package for comfyui_vllm_omni."""  # This is not a Python library intended to be imported.
 
 __all__ = [
     "NODE_CLASS_MAPPINGS",
@@ -13,9 +17,13 @@ __version__ = "0.0.1"
 from .comfyui_vllm_omni.nodes import (
     VLLMOmniARSampling,
     VLLMOmniDiffusionSampling,
+    VLLMOmniFastH3Deployment,
     VLLMOmniGenerateImage,
+    VLLMOmniGenerateMusic,
     VLLMOmniGenerateVideo,
+    VLLMOmniLatentMaskEditing,
     VLLMOmniMiniMaxH3Params,
+    VLLMOmniMiniMaxH3TemporalMask,
     VLLMOmniQwenTTSParams,
     VLLMOmniRemoteLoRA,
     VLLMOmniSamplingParamsList,
@@ -33,13 +41,17 @@ NODE_CLASS_MAPPINGS = {
     "VLLMOmniGenerateVideo": VLLMOmniGenerateVideo,
     "VLLMOmniUnderstanding": VLLMOmniUnderstanding,
     "VLLMOmniTTS": VLLMOmniTTS,
+    "VLLMOmniGenerateMusic": VLLMOmniGenerateMusic,
     "VLLMOmniVoiceClone": VLLMOmniVoiceClone,
     "VLLMOmniVideoReferences": VLLMOmniVideoReferences,
+    "VLLMOmniLatentMaskEditing": VLLMOmniLatentMaskEditing,
+    "VLLMOmniMiniMaxH3TemporalMask": VLLMOmniMiniMaxH3TemporalMask,
     # === Params ===
     "VLLMOmniARSampling": VLLMOmniARSampling,
     "VLLMOmniDiffusionSampling": VLLMOmniDiffusionSampling,
     "VLLMOmniSamplingParamsList": VLLMOmniSamplingParamsList,
     "VLLMOmniRemoteLoRA": VLLMOmniRemoteLoRA,
+    "VLLMOmniFastH3Deployment": VLLMOmniFastH3Deployment,
     "VLLMOmniQwenTTSParams": VLLMOmniQwenTTSParams,
     "VLLMOmniWanParams": VLLMOmniWanParams,
     "VLLMOmniMiniMaxH3Params": VLLMOmniMiniMaxH3Params,
@@ -52,13 +64,17 @@ NODE_DISPLAY_NAME_MAPPINGS = {
     "VLLMOmniGenerateVideo": "Generate Video",
     "VLLMOmniUnderstanding": "Multimodality Understanding",
     "VLLMOmniTTS": "TTS (Text to Speech)",
+    "VLLMOmniGenerateMusic": "Generate Music",
     "VLLMOmniVoiceClone": "TTS Voice Cloning",
     "VLLMOmniVideoReferences": "Video References",
+    "VLLMOmniLatentMaskEditing": "Latent Mask Editing",
+    "VLLMOmniMiniMaxH3TemporalMask": "MiniMax-H3 Temporal Mask",
     # === Params ===
     "VLLMOmniARSampling": "AR Sampling Params",
     "VLLMOmniDiffusionSampling": "Diffusion Sampling Params",
     "VLLMOmniSamplingParamsList": "Multi-Stage Sampling Params List",
     "VLLMOmniRemoteLoRA": "LoRA",
+    "VLLMOmniFastH3Deployment": "FastH3 Deployment",
     "VLLMOmniQwenTTSParams": "Qwen TTS Params",
     "VLLMOmniWanParams": "Wan Video Params",
     "VLLMOmniMiniMaxH3Params": "MiniMax-H3 Video Params",

apps/ComfyUI-vLLM-Omni/comfyui_vllm_omni/nodes.py

@@ -1,15 +1,25 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+
+import math
 from typing import Literal
 
 import torch
 from comfy_api.input import AudioInput, VideoInput
 
 from .utils.api_client import VLLMOmniClient
+from .utils.latent_mask import _align_frame_count
 from .utils.logger import get_logger
 from .utils.models import lookup_model_spec
 from .utils.types import (
+    MAX_REFERENCE_AUDIOS,
+    MAX_REFERENCE_IMAGES,
+    MAX_REFERENCE_VIDEOS,
     AudioFormat,
     AutoregressionSamplingParams,
     DiffusionSamplingParams,
+    FastH3Deployment,
+    LatentMaskEditing,
     MiniMaxH3ModelSpecificParams,
     QwenTTSModelSpecificParams,
     VideoReferences,
@@ -22,6 +32,23 @@ from .utils.validators import (
 
 logger = get_logger(__name__)
 
+FASTH3_INFERENCE_STEPS = 4
+FASTH3_FPS = 24
+# A FastH3 deployment may expose H3 under any --served-model-name, so the payload
+# spec has to be named outright rather than recovered from that alias.
+FASTH3_SPEC_MODEL = "MiniMax-H3"
+
+
+def _resolve_fast_h3_deployment(deployment: dict) -> tuple[str, str]:
+    if not isinstance(deployment, dict):
+        raise ValueError("FastH3 deployment must be provided by a FastH3 Deployment node.")
+
+    url = str(deployment.get("url") or "").strip().rstrip("/")
+    model = str(deployment.get("model") or "").strip()
+    if not url or not model:
+        raise ValueError("FastH3 deployment requires both URL and model.")
+    return url, model
+
 
 class _VLLMOmniGenerateBase:
     """Base class for vLLM-Omni generation nodes with shared functionality."""
@@ -162,14 +189,31 @@ class VLLMOmniGenerateVideo(_VLLMOmniGenerateBase):
                 "width": ("INT", {"default": 832, "min": 1}),
                 "height": ("INT", {"default": 480, "min": 1}),
                 "fps": ("INT", {"default": 16, "min": 1}),
-                "num_frames": ("INT", {"default": 41, "min": 1}),
+                "duration": (
+                    "FLOAT",
+                    {
+                        "default": 4.0,
+                        "min": 0.1,
+                        "step": 0.1,
+                        "round": 0.001,
+                        "tooltip": (
+                            "Clip length in seconds, converted to frames with the fps above. "
+                            "Models that only accept certain frame counts (e.g. MiniMax-H3) round to "
+                            "their own lattice, so the served clip can be slightly longer than requested."
+                        ),
+                    },
+                ),
             },
             "optional": {
                 "frame": ("IMAGE",),
+                "first_frame": ("IMAGE",),
+                "last_frame": ("IMAGE",),
                 "references": ("VIDEO_REFERENCES",),
                 "sampling_params": ("SAMPLING_PARAMS",),
                 "lora": ("REMOTE_LORA",),
                 "model_params": ("VIDEO_PARAMS",),
+                "fast_h3": ("FASTH3_DEPLOYMENT",),
+                "latent_edit": ("LATENT_MASK_EDITING",),
             },
         }
 
@@ -178,12 +222,30 @@ class VLLMOmniGenerateVideo(_VLLMOmniGenerateBase):
     FUNCTION = "generate"
 
     @classmethod
-    def VALIDATE_INPUTS(cls, url, model, frame=None, references=None, **_kwargs) -> str | Literal[True]:
+    def VALIDATE_INPUTS(
+        cls,
+        url,
+        model,
+        frame=None,
+        first_frame=None,
+        last_frame=None,
+        references=None,
+        fast_h3=None,
+        **_kwargs,
+    ) -> str | Literal[True]:
         base = super().VALIDATE_INPUTS(url, model)
         if base is not True:
             return base
         if frame is not None and references is not None:
             return "Provide only one of frame or references, not both."
+        if frame is not None and (first_frame is not None or last_frame is not None):
+            return "Provide either frame or first_frame/last_frame, not both."
+        if references is not None and (first_frame is not None or last_frame is not None):
+            return "Provide either first_frame/last_frame or references, not both."
+        if fast_h3 is not None and any(value is not None for value in (frame, first_frame, last_frame, references)):
+            return (
+                "FastH3 Preview supports T2VA only; disconnect frame, first_frame, last_frame, and references inputs."
+            )
         return True
 
     async def generate(
@@ -194,20 +256,72 @@ class VLLMOmniGenerateVideo(_VLLMOmniGenerateBase):
         width: int,
         height: int,
         fps: int,
-        num_frames: int,
+        duration: float,
         negative_prompt: str | None = None,
         frame: torch.Tensor | None = None,
+        first_frame: torch.Tensor | None = None,
+        last_frame: torch.Tensor | None = None,
         references: dict | None = None,
         sampling_params: dict | list[dict] | None = None,
         model_params: dict | None = None,
         lora: dict | None = None,
+        fast_h3: dict | None = None,
+        latent_edit: dict | None = None,
         **kwargs,
     ):
         if kwargs:
             logger.info("Uncaught kwargs: %s", kwargs)
         logger.debug("Got sampling params: %s", sampling_params)
         logger.debug("Got model params: %s", model_params)
-        validate_model_and_sampling_params_types(model, sampling_params)
+
+        # Which spec builds the payload. Only a FastH3 deployment separates this
+        # from the served name; every other path keeps them equal.
+        spec_model = model
+
+        if fast_h3 is not None:
+            if any(value is not None for value in (frame, first_frame, last_frame, references)):
+                raise ValueError(
+                    "FastH3 Preview supports T2VA only; disconnect frame, first_frame, "
+                    "last_frame, and references inputs."
+                )
+            if lora is not None:
+                raise ValueError(
+                    "FastH3 is already fused into the selected server; disconnect the request-level LoRA input."
+                )
+
+            url, model = _resolve_fast_h3_deployment(fast_h3)
+            logger.info("Using FastH3 deployment at %s", url)
+            fps = FASTH3_FPS
+            # The served name is whatever the operator passed to --served-model-name.
+            # Left alone, lookup_model_spec would miss H3 for an alias such as
+            # "fasth3", drop the params builder, and send a t2va request carrying no
+            # aspect_ratio -- which the server refuses.
+            spec_model = FASTH3_SPEC_MODEL
+
+            if sampling_params is None:
+                sampling_params = DiffusionSamplingParams()
+            elif isinstance(sampling_params, list):
+                if len(sampling_params) != 1:
+                    raise ValueError("FastH3 expects a single diffusion sampling params group.")
+                sampling_params = sampling_params[0].__class__(sampling_params[0])
+            else:
+                sampling_params = sampling_params.__class__(sampling_params)
+            sampling_params["num_inference_steps"] = FASTH3_INFERENCE_STEPS
+
+            # FastH3 owns both modality shifts. Sending the ordinary H3 values
+            # from a connected H3 Params node would turn a deployment choice
+            # into a request-level override, which the server intentionally
+            # rejects when it differs from the fused adapter contract.
+            if model_params is not None:
+                model_params = model_params.__class__(model_params)
+                model_params.pop("flow_shift", None)
+                model_params.pop("audio_flow_shift", None)
+
+        # Frames stay the wire unit. Convert after the FastH3 branch above, so the
+        # duration is measured against the fps the server will actually apply.
+        num_frames = max(1, round(duration * fps))
+
+        validate_model_and_sampling_params_types(spec_model, sampling_params)
 
         # Currently, all video generation models are single-stage diffusion models
         if isinstance(sampling_params, list):
@@ -227,8 +341,11 @@ class VLLMOmniGenerateVideo(_VLLMOmniGenerateBase):
         client = VLLMOmniClient(url)
         output = await client.generate_video(
             model=model,
+            spec_model=spec_model,
             prompt=prompt,
             frame=frame,  # frame present => fl2va / Wan I2V
+            first_frame=first_frame,
+            last_frame=last_frame,
             references=references,
             width=width,
             height=height,
@@ -238,6 +355,7 @@ class VLLMOmniGenerateVideo(_VLLMOmniGenerateBase):
             sampling_params=sampling_params,
             lora=lora,
             model_params=model_params,
+            latent_edit=latent_edit,
         )
         return (output,)
 
@@ -418,6 +536,66 @@ class VLLMOmniTTS(_VLLMOmniGenerateBase):
         return (audio,)
 
 
+class VLLMOmniGenerateMusic(_VLLMOmniGenerateBase):
+    """Generate a song from lyrics and a musical description with MiniMax Music 3."""
+
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {
+                "url": ("STRING", {"default": "http://localhost:8000/v1"}),
+                "model": ("STRING", {"default": "MiniMaxAI/MiniMax-Music3"}),
+                "instructions": (
+                    "STRING",
+                    {
+                        "multiline": True,
+                        "display_name": "caption",
+                        "tooltip": "Music caption: describe genre, instruments, tempo and mood.",
+                    },
+                ),
+                "lyrics": ("STRING", {"multiline": True}),
+                "max_duration_seconds": (
+                    "FLOAT",
+                    {
+                        "default": 300.0,
+                        "min": 1,
+                        "max": 360,
+                        "step": 0.01,
+                        "tooltip": "Upper limit; rounded down to whole 25 Hz audio frames. May end earlier.",
+                    },
+                ),
+                "seed": ("INT", {"default": 0, "min": 0, "max": 2**53 - 1, "control_after_generate": True}),
+                "response_format": (["wav", "mp3", "flac", "opus"],),
+            },
+        }
+
+    RETURN_TYPES = ("AUDIO",)
+    RETURN_NAMES = ("audio",)
+    FUNCTION = "generate"
+
+    async def generate(
+        self,
+        url: str,
+        model: str,
+        lyrics: str,
+        instructions: str,
+        response_format: AudioFormat,
+        max_duration_seconds: float,
+        seed: int = 0,
+    ) -> tuple[AudioInput]:
+        audio = await VLLMOmniClient(url.rstrip("/")).generate_speech(
+            model=model,
+            input=lyrics,
+            instructions=instructions,
+            voice="default",
+            speed=1.0,
+            response_format=response_format,
+            max_new_tokens=int(max_duration_seconds * 25),
+            seed=seed,
+        )
+        return (audio,)
+
+
 class VLLMOmniVoiceClone(_VLLMOmniGenerateBase):
     @classmethod
     def INPUT_TYPES(cls):
@@ -695,6 +873,52 @@ class VLLMOmniRemoteLoRA:
         return (lora,)
 
 
+class VLLMOmniFastH3Deployment:
+    """Select a vLLM-Omni service that fused FastH3 at startup."""
+
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {
+                "url": (
+                    "STRING",
+                    {
+                        "default": "http://localhost:8000/v1",
+                        "tooltip": "URL of a vLLM-Omni server started with a FastH3 --lora-path.",
+                    },
+                ),
+                "model": (
+                    "STRING",
+                    {
+                        "default": "MiniMaxAI/MiniMax-H3",
+                        "tooltip": "Model name exposed by the FastH3 deployment.",
+                    },
+                ),
+            }
+        }
+
+    RETURN_TYPES = ("FASTH3_DEPLOYMENT",)
+    RETURN_NAMES = ("deployment",)
+    FUNCTION = "get_deployment"
+    CATEGORY = "vLLM-Omni"
+    DESCRIPTION = (
+        "Selects a server with FastH3 fused at startup. The connected Generate Video node uses T2VA, "
+        "four inference steps, and 24 FPS without sending a request-level LoRA."
+    )
+
+    @classmethod
+    def VALIDATE_INPUTS(cls, url, model) -> str | Literal[True]:
+        try:
+            _resolve_fast_h3_deployment({"url": url, "model": model})
+        except ValueError as exc:
+            return str(exc)
+        return True
+
+    def get_deployment(self, url: str, model: str):
+        url, model = _resolve_fast_h3_deployment({"url": url, "model": model})
+        return (FastH3Deployment({"url": url, "model": model}),)
+
+
 class VLLMOmniQwenTTSParams:
     @classmethod
     def INPUT_TYPES(cls):
@@ -790,6 +1014,10 @@ class VLLMOmniVideoReferences:
                 "audio_2": ("AUDIO",),
                 "video_1": ("VIDEO",),
                 "video_2": ("VIDEO",),
+                # Append ports to preserve connections in saved workflows.
+                **{f"image_{i}": ("IMAGE",) for i in range(3, MAX_REFERENCE_IMAGES + 1)},
+                **{f"audio_{i}": ("AUDIO",) for i in range(3, MAX_REFERENCE_AUDIOS + 1)},
+                **{f"video_{i}": ("VIDEO",) for i in range(3, MAX_REFERENCE_VIDEOS + 1)},
             },
         }
 
@@ -806,21 +1034,112 @@ class VLLMOmniVideoReferences:
         audio_2: AudioInput | None = None,
         video_1: VideoInput | None = None,
         video_2: VideoInput | None = None,
+        image_3: torch.Tensor | None = None,
+        image_4: torch.Tensor | None = None,
+        image_5: torch.Tensor | None = None,
+        image_6: torch.Tensor | None = None,
+        image_7: torch.Tensor | None = None,
+        image_8: torch.Tensor | None = None,
+        image_9: torch.Tensor | None = None,
+        audio_3: AudioInput | None = None,
+        video_3: VideoInput | None = None,
         **kwargs,
     ):
         if kwargs:
             logger.info("Uncaught kwargs: %s", kwargs)
         refs = VideoReferences()
-        if image_1 is not None:
-            refs["image_1"] = image_1
-        if image_2 is not None:
-            refs["image_2"] = image_2
-        if audio_1 is not None:
-            refs["audio_1"] = audio_1
-        if audio_2 is not None:
-            refs["audio_2"] = audio_2
-        if video_1 is not None:
-            refs["video_1"] = video_1
-        if video_2 is not None:
-            refs["video_2"] = video_2
+        for kind, values in (
+            ("image", (image_1, image_2, image_3, image_4, image_5, image_6, image_7, image_8, image_9)),
+            ("video", (video_1, video_2, video_3)),
+            ("audio", (audio_1, audio_2, audio_3)),
+        ):
+            for index, value in enumerate(values, start=1):
+                if value is not None:
+                    refs[f"{kind}_{index}"] = value
         return (refs,)
+
+
+class VLLMOmniLatentMaskEditing:
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {},
+            "optional": {
+                "source_video": ("VIDEO",),
+                "source_audio": ("AUDIO",),
+                "video_mask": ("MASK",),
+                "audio_mask": ("FLOAT", {"default": -1.0, "min": -1.0, "max": 1.0, "step": 0.01}),
+            },
+        }
+
+    RETURN_TYPES = ("LATENT_MASK_EDITING",)
+    RETURN_NAMES = ("latent_edit",)
+    FUNCTION = "get_latent_edit"
+    CATEGORY = "vLLM-Omni"
+
+    def get_latent_edit(
+        self,
+        source_video: VideoInput | None = None,
+        source_audio: AudioInput | None = None,
+        video_mask: torch.Tensor | None = None,
+        audio_mask: float = -1.0,
+        **kwargs,
+    ):
+        if kwargs:
+            logger.info("Uncaught kwargs: %s", kwargs)
+        edit = LatentMaskEditing()
+        if source_video is not None:
+            edit["source_video"] = source_video
+        if source_audio is not None:
+            edit["source_audio"] = source_audio
+        if video_mask is not None:
+            edit["video_mask"] = video_mask
+        if audio_mask >= 0.0:
+            edit["audio_mask"] = audio_mask
+        return (edit,)
+
+
+class VLLMOmniMiniMaxH3TemporalMask:
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {
+                "images": ("IMAGE",),
+                "source_fps": ("FLOAT", {"default": 24.0, "min": 0.01}),
+                "duration": ("FLOAT", {"default": 5.0, "min": 0.01, "max": 15.0}),
+                "mode": (["continuation", "extension"],),
+                "preserve_fraction": ("FLOAT", {"default": 0.5, "min": 0.0, "max": 1.0}),
+            }
+        }
+
+    RETURN_TYPES = ("MASK", "FLOAT", "IMAGE", "MASK")
+    RETURN_NAMES = ("mask", "preview_fps", "preview_images", "preview_mask")
+    FUNCTION = "build"
+    CATEGORY = "vLLM-Omni"
+
+    def build(self, images, source_fps, duration, mode, preserve_fraction=0.5):
+        if images.shape[0] <= 0 or not math.isfinite(source_fps) or source_fps <= 0:
+            raise ValueError("Source must contain frames and have a positive finite FPS.")
+        if not math.isfinite(duration) or duration <= 0 or not 0 <= preserve_fraction <= 1:
+            raise ValueError("Invalid duration or preserve_fraction.")
+        frames = _align_frame_count(max(1, round(duration * 24)))
+        source_seconds = images.shape[0] / source_fps
+        if mode == "extension":
+            if frames / 24 <= source_seconds:
+                raise ValueError("Extension output must be longer than the source; increase duration.")
+            boundary = source_seconds
+        elif mode == "continuation":
+            boundary = min(source_seconds, frames / 24) * preserve_fraction
+        else:
+            raise ValueError(f"Unknown temporal mask mode: {mode}")
+        available = min(frames, math.floor(boundary * 24 + 1e-8))
+        # Snap the preserved prefix to whole VAE clips (5 + 17n frames) so every
+        # latent it maps to is fully preserved.
+        prefix = 0 if available < 5 else 5 + 17 * ((available - 5) // 17)
+        # One slice per output frame; the server pools frames to the latent grid.
+        mask = torch.ones(frames, 1, 1)
+        mask[:prefix] = 0
+        indices = torch.arange(frames, device=images.device)
+        source_indices = (indices * (source_fps / 24)).floor().long().clamp(max=images.shape[0] - 1)
+        preview_images = images.index_select(0, source_indices)
+        return mask, 24.0, preview_images, mask

apps/ComfyUI-vLLM-Omni/comfyui_vllm_omni/utils/api_client.py

@@ -1,3 +1,6 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+
 """
 An high-level API client adapter that forwards ComfyUI inputs to vLLM-Omni's REST API,
 and transforms the API responses back to ComfyUI formats.
@@ -17,6 +20,7 @@ from comfy_api.input import AudioInput, VideoInput
 
 from .format import (
     audio_to_base64,
+    audio_to_bytes,
     base64_to_audio,
     base64_to_image_tensor,
     bytes_to_audio,
@@ -26,9 +30,16 @@ from .format import (
     video_to_base64,
     video_to_bytes,
 )
+from .latent_mask import scalar_mask_to_json, video_mask_to_json
 from .logger import get_logger, pretty_printer
 from .models import lookup_model_spec
-from .types import AudioFormat
+from .types import (
+    MAX_REFERENCE_AUDIOS,
+    MAX_REFERENCE_IMAGES,
+    MAX_REFERENCE_VIDEOS,
+    MAX_TOTAL_REFERENCES,
+    AudioFormat,
+)
 
 logger = get_logger(__name__)
 
@@ -267,14 +278,34 @@ class VLLMOmniClient:
         fps: int,
         negative_prompt: str | None = None,
         frame: torch.Tensor | None = None,
+        first_frame: torch.Tensor | None = None,
+        last_frame: torch.Tensor | None = None,
         references: dict | None = None,
         sampling_params: dict | None = None,
         model_params: dict | None = None,
         lora: dict | None = None,
+        latent_edit: dict | None = None,
+        spec_model: str | None = None,
         **extra_params,
     ) -> VideoInput:
+        """Post a video job and return the decoded result.
+
+        ``spec_model`` names the model whose payload spec builds the request, for
+        deployments that serve a known model under a different ``model`` alias. It
+        never reaches the wire; defaults to ``model``.
+        """
         if frame is not None and references is not None:
             raise ValueError("Provide only one of frame or references, not both.")
+        if frame is not None and (first_frame is not None or last_frame is not None):
+            raise ValueError("Provide either frame or first_frame/last_frame, not both.")
+        if references is not None and (first_frame is not None or last_frame is not None):
+            raise ValueError("Provide either first_frame/last_frame or references, not both.")
+
+        spec, matched_pattern = lookup_model_spec(spec_model or model)
+        if (first_frame is not None or last_frame is not None) and (
+            matched_pattern is None or "MiniMax-H3" not in matched_pattern
+        ):
+            raise ValueError("first_frame and last_frame are supported only for MiniMax-H3; use frame for this model.")
 
         # === regular payload fields ===
         form = aiohttp.FormData()
@@ -294,43 +325,62 @@ class VLLMOmniClient:
 
         # === multimodal inputs (first-last-frames, references, etc.) ===
         input_reference_image: torch.Tensor | None = None
-        audio_reference: AudioInput | None = None
-        reference_videos: list[VideoInput] = []
+        keyframe_images: list[tuple[str, torch.Tensor]] = []
         video_task: str | None = None
 
         if frame is not None:
             input_reference_image = frame
             video_task = "fl2va"
+        elif first_frame is not None or last_frame is not None:
+            frame_indices: list[int] = []
+            if first_frame is not None:
+                keyframe_images.append(("first_frame.png", first_frame))
+                frame_indices.append(0)
+            if last_frame is not None:
+                keyframe_images.append(("last_frame.png", last_frame))
+                frame_indices.append(-1)
+            if len(keyframe_images) == 1:
+                input_reference_image = keyframe_images[0][1]
+            extra_params["frame_indices"] = frame_indices
+            video_task = "fl2va"
         elif references is not None:
-            images = [references[k] for k in ("image_1", "image_2") if k in references and references[k] is not None]
-            audios = [references[k] for k in ("audio_1", "audio_2") if k in references and references[k] is not None]
-            videos = [references[k] for k in ("video_1", "video_2") if k in references and references[k] is not None]
-
-            if not images and not audios and not videos:
-                raise ValueError("references is empty; connect at least one image/audio/video.")
-
-            if videos:
-                if images or audios:
-                    raise ValueError(
-                        "Video references cannot be combined with image or audio references. "
-                        "Connect only video_1/video_2 for multi-video Ref2VA."
-                    )
-                reference_videos = videos
-                video_task = "ref2va"
-            elif len(images) == 1 and len(audios) == 1:
-                input_reference_image = images[0]
-                audio_reference = audios[0]
-                video_task = "ref2va"
-            else:
+            reference_formats = (
+                ("image", MAX_REFERENCE_IMAGES, "png", "image/png", image_tensor_to_png_bytes),
+                ("video", MAX_REFERENCE_VIDEOS, "mp4", "video/mp4", video_to_bytes),
+                ("audio", MAX_REFERENCE_AUDIOS, "mp3", "audio/mpeg", audio_to_bytes),
+            )
+            supported_inputs = {f"{kind}_{i}" for kind, limit, *_ in reference_formats for i in range(1, limit + 1)}
+            connected = {name: value for name, value in references.items() if value is not None}
+            unsupported = connected.keys() - supported_inputs
+            if unsupported:
+                raise ValueError(f"Unsupported reference input(s): {', '.join(sorted(unsupported))}.")
+            if not any(name.startswith(("image_", "video_")) for name in connected):
                 raise ValueError(
-                    "Invalid references combination. Supported modes: "
-                    "(1) one or more videos only, or (2) exactly one image and one audio."
+                    "references requires at least one image or video; audio-only inputs are not supported."
                 )
+            if len(connected) > MAX_TOTAL_REFERENCES:
+                raise ValueError(
+                    f"references supports at most {MAX_TOTAL_REFERENCES} inputs in total "
+                    f"(up to {MAX_REFERENCE_IMAGES} images, {MAX_REFERENCE_VIDEOS} videos, "
+                    f"and {MAX_REFERENCE_AUDIOS} audios)."
+                )
+            for kind, limit, extension, content_type, encode in reference_formats:
+                for index in range(1, limit + 1):
+                    name = f"{kind}_{index}"
+                    if name in connected:
+                        filename = f"{name}.{extension}"
+                        form.add_field(
+                            "input_references",
+                            encode(connected[name], filename),
+                            filename=filename,
+                            content_type=content_type,
+                        )
+            video_task = "ref2va"
         else:
             video_task = "t2va"
 
         if input_reference_image is not None:
-            image_filename = "image.png"  # Required for multipart form
+            image_filename = keyframe_images[0][0] if keyframe_images else "image.png"
             form.add_field(
                 "input_reference",
                 image_tensor_to_png_bytes(input_reference_image, image_filename),
@@ -338,32 +388,69 @@ class VLLMOmniClient:
                 content_type="image/png",
             )
 
-        if audio_reference is not None:
-            form.add_field(
-                "audio_reference",
-                json.dumps({"audio_url": audio_to_base64(audio_reference)}, ensure_ascii=False),
-            )
+        if len(keyframe_images) == 2:
+            for image_filename, image in keyframe_images:
+                form.add_field(
+                    "input_references",
+                    image_tensor_to_png_bytes(image, image_filename),
+                    filename=image_filename,
+                    content_type="image/png",
+                )
 
-        for idx, video in enumerate(reference_videos, start=1):
-            video_filename = f"reference_{idx}.mp4"
-            form.add_field(
-                "input_references",
-                video_to_bytes(video, video_filename),
-                filename=video_filename,
-                content_type="video/mp4",
-            )
+        # === latent-mask editing (MiniMax H3) ===
+        if latent_edit is not None:
+            source_video = latent_edit.get("source_video")
+            source_audio = latent_edit.get("source_audio")
+            video_mask = latent_edit.get("video_mask")
+            audio_mask = latent_edit.get("audio_mask")
+
+            if video_mask is None and audio_mask is None:
+                raise ValueError("Latent-mask editing requires at least one mask.")
+
+            if source_video is not None:
+                form.add_field(
+                    "source_video",
+                    video_to_bytes(source_video),
+                    filename="source.mp4",
+                    content_type="video/mp4",
+                )
+            if source_audio is not None:
+                # The filename extension selects the audio codec (mp3), so it is
+                # meaningful even though aiohttp uses the explicit filename below.
+                form.add_field(
+                    "source_audio",
+                    audio_to_bytes(source_audio, "source_audio.mp3"),
+                    filename="source_audio.mp3",
+                    content_type="audio/mpeg",
+                )
+            if video_mask is not None:
+                mask_json = video_mask_to_json(video_mask)
+                form.add_field(
+                    "video_noise_mask",
+                    mask_json.encode("utf-8"),
+                    filename="video-mask.json",
+                    content_type="application/json",
+                )
+            if audio_mask is not None:
+                form.add_field(
+                    "audio_noise_mask",
+                    scalar_mask_to_json(audio_mask).encode("utf-8"),
+                    filename="audio-mask.json",
+                    content_type="application/json",
+                )
 
         # === model specific params. Either use a specialized builder, or add flattened fields as-is ===
         if model_params is not None:
             model_params = dict(model_params)
             model_params.pop("type", None)
 
-        spec, _ = lookup_model_spec(model)
         params_builder = spec.get("params_builder") if spec else None
         if params_builder is not None:
             form_fields = params_builder(
                 model_params or {},
                 extra_params={**extra_params, "task": video_task},
+                width=width,
+                height=height,
             )
             for k, v in form_fields.items():
                 form.add_field(k, v if isinstance(v, str) else str(v))

apps/ComfyUI-vLLM-Omni/comfyui_vllm_omni/utils/format.py

@@ -1,3 +1,6 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+
 """Image/tensor format helpers.
 
 The image generation part is derived from dougbtv/comfyui-vllm-omni by Doug (@dougbtv).
@@ -170,6 +173,10 @@ def bytes_to_video(video_bytes: bytes) -> VideoInput:
                 audio_stream = container.streams.audio[-1]
                 audio_frames = []
                 resampler = AudioResampler(format="fltp")
+                # Decoding the video stream above left the container at EOF.
+                # Rewind before demuxing again, or this pass yields no frames
+                # and the audio of a T2VA result is silently dropped.
+                container.seek(0)
                 for frame in container.decode(audio_stream):
                     if not isinstance(frame, AudioFrame):
                         continue

apps/ComfyUI-vLLM-Omni/comfyui_vllm_omni/utils/latent_mask.py

@@ -0,0 +1,57 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+
+"""Client-side serialization of MiniMax-H3 latent-edit masks.
+
+A video mask is sent in frame space — a 2D spatial ``[H, W]`` (applied to every
+frame) or a 3D ``[T, H, W]`` (one slice per frame) — and the server resolves it
+to the latent grid. An audio mask is a scalar applied to all time steps.
+"""
+
+import json
+import math
+
+import torch
+import torch.nn.functional as F
+
+# The server rejects mask JSON above 8 MiB, so the spatial axes are
+# area-downsampled by the VAE spatial stride before upload. This depends only on
+# the stride, not on the H3 shape lattice, which the server still owns.
+_VAE_SPATIAL_STRIDE = 16
+_MASK_DECIMALS = 4
+
+
+def scalar_mask_to_json(value: float) -> str:
+    if not 0.0 <= value <= 1.0:
+        raise ValueError(f"mask value must be in [0, 1], got {value}")
+    return str(value)
+
+
+def video_mask_to_json(mask: torch.Tensor) -> str:
+    """Serialize a frame-space video mask to JSON for the ``video_noise_mask`` field.
+
+    ``mask`` is a 2D spatial mask or a 3D frame-space mask. Its spatial axes are
+    area-downsampled by the VAE stride; the server resizes the result to the
+    latent grid, so the client neither floors the canvas nor maps frames to
+    latents.
+    """
+    if mask.ndim not in (2, 3):
+        raise ValueError(f"expected a 2D or 3D mask tensor, got {mask.ndim}D")
+    frames = mask.unsqueeze(0) if mask.ndim == 2 else mask
+    size = tuple(max(1, math.ceil(dim / _VAE_SPATIAL_STRIDE)) for dim in frames.shape[1:])
+    frames = F.interpolate(frames.unsqueeze(1).float(), size=size, mode="area").squeeze(1)
+    if mask.ndim == 2:
+        frames = frames.squeeze(0)
+    # Round in float64 so tolist() emits short decimals rather than float32 noise.
+    return json.dumps(frames.double().round(decimals=_MASK_DECIMALS).tolist(), separators=(",", ":"))
+
+
+# The temporal-mask node snaps its preserve boundary to the H3 frame lattice;
+# this mirrors the server shape planner.
+def _align_frame_count(frame_count: int) -> int:
+    if frame_count <= 0:
+        return 1
+    current = int(frame_count)
+    while current % 17 != 5:
+        current += 1
+    return current

apps/ComfyUI-vLLM-Omni/comfyui_vllm_omni/utils/models.py

@@ -1,3 +1,6 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+
 import json
 import re
 from typing import Any
@@ -34,10 +37,28 @@ def _qwen25_payload_preprocessor(payload: dict) -> dict:
     return payload
 
 
+MINIMAX_H3_ASPECT_RATIOS = {
+    "21:9": 21.0 / 9.0,
+    "16:9": 16.0 / 9.0,
+    "4:3": 4.0 / 3.0,
+    "1:1": 1.0,
+    "3:4": 3.0 / 4.0,
+    "9:16": 9.0 / 16.0,
+}
+
+
+def _nearest_minimaxh3_aspect_ratio(width: int, height: int) -> str:
+    """Pick the supported named ratio closest to the requested frame size."""
+    target = float(width) / float(height)
+    return min(MINIMAX_H3_ASPECT_RATIOS, key=lambda name: abs(MINIMAX_H3_ASPECT_RATIOS[name] - target))
+
+
 def _minimaxh3_params_builder(
     model_params: dict[str, Any],
     *,
     extra_params: dict[str, Any],
+    width: int | None = None,
+    height: int | None = None,
 ) -> dict[str, Any]:
     """Build multipart form fields for MiniMax-H3 from model_params + routed task."""
     params = dict(model_params)
@@ -53,6 +74,12 @@ def _minimaxh3_params_builder(
     if params:
         logger.warning("Unused MiniMax-H3 model params ignored: %s", sorted(params))
 
+    # H3 refuses t2va without a named aspect ratio, and the Generate Video node
+    # carries width/height rather than a ratio. fl2va takes its ratio from the
+    # input image and ref2va defaults server-side, so only t2va needs this.
+    if merged_extra_params.get("task") == "t2va" and "aspect_ratio" not in merged_extra_params and width and height:
+        merged_extra_params["aspect_ratio"] = _nearest_minimaxh3_aspect_ratio(width, height)
+
     if merged_extra_params:
         form_fields["extra_params"] = json.dumps(merged_extra_params, ensure_ascii=False)
     return form_fields

apps/ComfyUI-vLLM-Omni/comfyui_vllm_omni/utils/types.py

@@ -1,3 +1,6 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+
 from collections.abc import Callable
 from enum import Enum, auto
 from typing import Any, Literal, TypeAlias
@@ -27,10 +30,24 @@ class MiniMaxH3ModelSpecificParams(dict):
     pass
 
 
+MAX_REFERENCE_IMAGES = 9
+MAX_REFERENCE_VIDEOS = 3
+MAX_REFERENCE_AUDIOS = 3
+MAX_TOTAL_REFERENCES = 12
+
+
 class VideoReferences(dict):
     pass
 
 
+class LatentMaskEditing(dict):
+    pass
+
+
+class FastH3Deployment(dict):
+    """Descriptor for a server that fused FastH3 at startup."""
+
+
 class ModelMode(Enum):
     IMAGE_GENERATION = auto()
     VIDEO_GENERATION = auto()

apps/ComfyUI-vLLM-Omni/docs/minimax-h3-t2v.md

@@ -0,0 +1,94 @@
+# MiniMax H3 Text to Video (WF-01)
+
+Import [MiniMax_H3_Text_to_Video.json](../example_workflows/MiniMax_H3_Text_to_Video.json)
+into ComfyUI, or select it under **Templates → ComfyUI-vLLM-Omni**.
+The template connects existing nodes to a remote H3 service and saves the
+returned video with audio. ComfyUI does not load H3 weights.
+
+## Setup
+
+Install the [ComfyUI extension](../README.md#installation) and follow the
+[H3 deployment recipe](../../../recipes/MiniMaxAI/MiniMax-H3.md) for your hardware.
+For a two-GPU FL2VA server with sufficient memory:
+
+```bash
+export MODEL=/path/to/MiniMax-H3/FL2VA
+CUDA_VISIBLE_DEVICES=0,1 VLLM_WORKER_MULTIPROC_METHOD=spawn \
+VLLM_OMNI_VIDEO_SYNC_TIMEOUT=14400 \
+vllm serve "$MODEL" --omni --host 127.0.0.1 --port 8000 \
+  --served-model-name MiniMaxAI/MiniMax-H3 --trust-remote-code \
+  --task-type fl2va --num-gpus 2 --tensor-parallel-size 2 \
+  --text-encoder-tp-size 2 --enforce-eager
+```
+
+Set Generate Video's **url** to the address reachable from ComfyUI, including
+`/v1`, and match its **model** to the served name. The defaults are
+`http://localhost:8000/v1` and `MiniMaxAI/MiniMax-H3`.
+
+## Base and Turbo
+
+Base presets are connected by default. Write the scene and its sound in the
+same **prompt**; leave **frame**, **references**, and **fast_h3** disconnected.
+The template uses 1344×768, 24 FPS, seed 1101 and **duration** 5.167 seconds
+(124 frames, satisfying H3's `17k+5` constraint). The existing client infers
+`16:9` from these dimensions. For portrait output, use 768×1344.
+
+| Setting | Base | Turbo v1.0 768p |
+| --- | --- | --- |
+| Inference steps | 50 | 5 (4 forwards) |
+| Video flow shift | 12 | 6 |
+| Audio flow shift | 3 | 3 |
+| Remote LoRA | Disconnected | Connected, scale 1 |
+
+For Turbo, use the **Diffusers-layout** artifact from
+[lightx2v/Minimax-h3-Turbo](https://huggingface.co/lightx2v/Minimax-h3-Turbo):
+
+```text
+minimax_h3_fl2v_turbo_4step_v1.0_768p_bf16.safetensors
+```
+
+Add `--lora-backend peft --lora-path "$TURBO_LORA"` to the server command,
+where `TURBO_LORA` is the full server-side path to that file. Enter the same
+path in Remote LoRA's **local_path**; retain name `h3-turbo-v1.0-768p`, scale
+`1.0`, and **int_id** `0` so the existing API derives the ID from the path.
+The template leaves the path blank for users to configure. The `_comfyui_`
+export is not supported by this backend.
+
+Connect **Turbo sampling**, **Turbo H3 params**, and **Remote LoRA** to Generate
+Video's `sampling_params`, `model_params`, and `lora` inputs, replacing the
+Base links. To return to Base, restore both Base links and disconnect LoRA.
+Other Turbo artifacts may need different settings; see the
+[artifact table](../../../recipes/MiniMaxAI/MiniMax-H3.md#turbo-lora).
+
+If importing an older template with a frame-count widget, reimport this JSON
+or set **duration** to **5.167**; do not interpret 124 frames as 124 seconds.
+
+## Validation
+
+Queue the configured workflow and inspect the MP4 saved under
+`ComfyUI/output/video/MiniMax_H3_WF01_*.mp4`:
+
+```bash
+ffprobe -v error -show_streams -show_format -of json "$OUTPUT"
+ffmpeg -v error -i "$OUTPUT" -f null -
+```
+
+Check the requested dimensions, frame count, 24 FPS and both video/audio
+streams, then watch and listen to the saved file. The current upstream
+extension includes the shared audio-preserving decoder fix.
+
+On September 16, the narrowed workflow was validated through ComfyUI using
+existing upstream Python code at `4a8ac297`, two H20-3e GPUs, vLLM 0.29.0,
+PyTorch 2.13.0+cu129 and ComfyUI `1d48d9cf`. The run used the template's
+forest-stream prompt, explicit server-side Turbo LoRA path, and settings
+above, with `TORCH_SDPA`, `--usp 1 --ring 1`, and tiled VAE parallelism on two
+GPUs. ComfyUI execution took 131.55 seconds. SaveVideo produced 124 H.264
+frames at 1344×768 and 24 FPS, plus 32 kHz stereo AAC audio. Video/audio
+durations were 5.166667/5.167000 seconds; full decoding passed and audio was
+nonzero. This is model-generated output, not a ground-truth reference or a
+subjective audio-quality assessment. Base graph validation passed with Turbo
+nodes disconnected; full 50-step Base quality was not evaluated.
+
+Based on the official
+[ComfyUI H3 T2V template](https://github.com/Comfy-Org/workflow_templates/blob/main/templates/video_minimax_h3_t2v.json),
+using remote inference instead of local model loaders.

apps/ComfyUI-vLLM-Omni/docs/wf05-h3-latent-editing.md

@@ -0,0 +1,17 @@
+# WF-05: MiniMax-H3 latent-mask editing
+
+Edit a source video with MiniMax-H3 using spatial or temporal masks. The workflow includes object removal, inpainting, continuation, and extension, with a mask preview for each case.
+
+## Requirements
+
+- ComfyUI with ComfyUI-vLLM-Omni installed.
+- An H3 service using the FL2VA checkpoint partition with latent-mask editing support. Follow the [H3 recipe](../../../recipes/MiniMaxAI/MiniMax-H3.md) for server setup.
+
+## Use
+
+Open **vLLM-Omni MiniMax-H3 Latent Mask Editing.json** from the example workflows and set the Generate Video URL/model for the case you want to run.
+
+1. Upload your source video in Load Video.
+2. Adjust the case's mask. For removal and inpainting, set Mask Canvas to the video dimensions, Mask Rectangle to the region size, and Combine Masks x/y to its position, or replace that chain with Load Image (as Mask) for an arbitrary-shape mask. For continuation and extension, set MiniMax-H3 Temporal Mask's mode and duration; continuation also uses `preserve_fraction`. Match the duration in Generate Video, and use a duration longer than the source for extension. Mask values are `0` to preserve and `1` to regenerate.
+3. Adjust the prompt to describe the desired content. Set the output dimensions and duration in Generate Video. If editing audio, update the sound description and audio mask (`0` preserves, `1` regenerates the whole clip).
+4. Select the case's Mask Preview Save Video node and choose **Execute to selected output nodes** to preview the mask as a red overlay without inference, or select its Result Save Video node to generate the edited video. Running the entire workflow generates all four cases.

apps/ComfyUI-vLLM-Omni/docs/wf07-h3-upscale.md

@@ -0,0 +1,21 @@
+# WF-07: H3 video upscale with SeedVR2
+
+Generate video and audio through vLLM-Omni H3, upscale the decoded frames with SeedVR2, and save both the original and upscaled videos. The generated audio and FPS connect directly to Create Video.
+
+## Requirements
+
+- ComfyUI with ComfyUI-vLLM-Omni and [ComfyUI-SeedVR2_VideoUpscaler](https://github.com/numz/ComfyUI-SeedVR2_VideoUpscaler#-installation) installed.
+- An H3 service using the FL2VA checkpoint partition. Follow the [H3 recipe](../../../recipes/MiniMaxAI/MiniMax-H3.md) for server setup.
+- `seedvr2_ema_3b_fp16.safetensors` and `ema_vae_fp16.safetensors` from [SeedVR2's ComfyUI weights](https://huggingface.co/numz/SeedVR2_comfyUI/tree/main), placed in `ComfyUI/models/SEEDVR2`. The SeedVR2 nodes can also download them on first use.
+
+## Use
+
+Open **vLLM-Omni MiniMax H3 Video Upscale.json** from the example workflows. An API-format version is provided alongside it. Set the Generate Video URL/model and the LoRA path as seen by the H3 server.
+
+The preset uses the LightX2V v1.0 768p Turbo adapter in Diffusers format: `minimax_h3_fl2v_turbo_4step_v1.0_768p_bf16.safetensors`, five sigma points, video/audio flow shifts 6/3, and LoRA scale 1. Preload the same adapter on the server using the [Turbo LoRA recipe](../../../recipes/MiniMaxAI/MiniMax-H3.md#turbo-lora). For base H3, disconnect LoRA, set inference steps to 50, and set video flow shift to 12.
+
+The H3 preset requests 1344 x 768 at 24 FPS for 5.167 seconds, which gives 124 frames. SeedVR2's `resolution=1536` sets the target short edge, producing 2688 x 1536 for this input. Update this value when changing the input size; it is an output resolution, not a fixed 2x multiplier.
+
+The SeedVR2 preset uses 3B FP16, five-frame batches, one overlapping frame, uniform batch padding, and LAB color correction. The DiT loader uses CPU offload with 32 swapped blocks; the VAE uses tiled encoding and decoding. Select the local GPU in both model loaders. See the [SeedVR2 settings](https://github.com/numz/ComfyUI-SeedVR2_VideoUpscaler#-usage) when adjusting memory use or batch size.
+
+Run the workflow. The two Save Video nodes write MP4s under `WF07/generated` and `WF07/upscaled` in ComfyUI's output directory.

apps/ComfyUI-vLLM-Omni/example_workflows/MiniMax_H3_Text_to_Video.json

@@ -0,0 +1,362 @@
+{
+  "last_node_id": 8,
+  "last_link_id": 3,
+  "nodes": [
+    {
+      "id": 1,
+      "type": "VLLMOmniDiffusionSampling",
+      "pos": [
+        20,
+        20
+      ],
+      "size": [
+        310,
+        280
+      ],
+      "flags": {},
+      "order": 0,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "diffusion sampling params",
+          "type": "SAMPLING_PARAMS",
+          "links": [
+            1
+          ],
+          "slot_index": 0
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniDiffusionSampling"
+      },
+      "widgets_values": [
+        1,
+        50,
+        1.0,
+        1.0,
+        false,
+        false,
+        1101,
+        "fixed"
+      ],
+      "title": "Base sampling"
+    },
+    {
+      "id": 2,
+      "type": "VLLMOmniMiniMaxH3Params",
+      "pos": [
+        20,
+        340
+      ],
+      "size": [
+        310,
+        140
+      ],
+      "flags": {},
+      "order": 1,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "MiniMax-H3 video params",
+          "type": "VIDEO_PARAMS",
+          "links": [
+            2
+          ],
+          "slot_index": 0
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniMiniMaxH3Params"
+      },
+      "widgets_values": [
+        3.0,
+        12.0
+      ],
+      "title": "Base H3 params"
+    },
+    {
+      "id": 3,
+      "type": "VLLMOmniGenerateVideo",
+      "pos": [
+        430,
+        20
+      ],
+      "size": [
+        430,
+        440
+      ],
+      "flags": {},
+      "order": 2,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "frame",
+          "type": "IMAGE",
+          "link": null
+        },
+        {
+          "name": "references",
+          "type": "VIDEO_REFERENCES",
+          "link": null
+        },
+        {
+          "name": "sampling_params",
+          "type": "SAMPLING_PARAMS",
+          "link": 1
+        },
+        {
+          "name": "lora",
+          "type": "REMOTE_LORA",
+          "link": null
+        },
+        {
+          "name": "model_params",
+          "type": "VIDEO_PARAMS",
+          "link": 2
+        },
+        {
+          "name": "fast_h3",
+          "type": "FASTH3_DEPLOYMENT",
+          "link": null
+        },
+        {
+          "name": "duration",
+          "type": "FLOAT",
+          "widget": {
+            "name": "duration"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "name": "video",
+          "type": "VIDEO",
+          "links": [
+            3
+          ],
+          "slot_index": 0
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniGenerateVideo"
+      },
+      "widgets_values": [
+        "http://localhost:8000/v1",
+        "MiniMaxAI/MiniMax-H3",
+        "A small stream flows through a sunlit forest. The camera slowly moves toward the water. Birds chirp in the trees and water gurgles over smooth stones. Natural continuous sound, no music.",
+        "",
+        1344,
+        768,
+        24,
+        5.167
+      ]
+    },
+    {
+      "id": 4,
+      "type": "SaveVideo",
+      "pos": [
+        940,
+        20
+      ],
+      "size": [
+        440,
+        440
+      ],
+      "flags": {},
+      "order": 3,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "video",
+          "type": "VIDEO",
+          "link": 3
+        }
+      ],
+      "outputs": [
+        {
+          "name": "video",
+          "type": "VIDEO",
+          "links": null,
+          "slot_index": 0
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "SaveVideo"
+      },
+      "widgets_values": [
+        "video/MiniMax_H3_WF01",
+        "mp4",
+        "auto"
+      ]
+    },
+    {
+      "id": 5,
+      "type": "VLLMOmniRemoteLoRA",
+      "pos": [
+        430,
+        540
+      ],
+      "size": [
+        430,
+        160
+      ],
+      "flags": {},
+      "order": 4,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "lora",
+          "type": "REMOTE_LORA",
+          "links": null,
+          "slot_index": 0
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniRemoteLoRA"
+      },
+      "widgets_values": [
+        "",
+        "h3-turbo-v1.0-768p",
+        1.0,
+        0
+      ],
+      "title": "Optional Turbo: server LoRA path"
+    },
+    {
+      "id": 6,
+      "type": "VLLMOmniDiffusionSampling",
+      "pos": [
+        20,
+        540
+      ],
+      "size": [
+        310,
+        280
+      ],
+      "flags": {},
+      "order": 5,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "diffusion sampling params",
+          "type": "SAMPLING_PARAMS",
+          "links": null,
+          "slot_index": 0
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniDiffusionSampling"
+      },
+      "widgets_values": [
+        1,
+        5,
+        1.0,
+        1.0,
+        false,
+        false,
+        1101,
+        "fixed"
+      ],
+      "title": "Turbo sampling (4 forwards)"
+    },
+    {
+      "id": 7,
+      "type": "VLLMOmniMiniMaxH3Params",
+      "pos": [
+        430,
+        740
+      ],
+      "size": [
+        310,
+        140
+      ],
+      "flags": {},
+      "order": 6,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "MiniMax-H3 video params",
+          "type": "VIDEO_PARAMS",
+          "links": null,
+          "slot_index": 0
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniMiniMaxH3Params"
+      },
+      "widgets_values": [
+        3.0,
+        6.0
+      ],
+      "title": "Turbo H3 params"
+    },
+    {
+      "id": 8,
+      "type": "Note",
+      "pos": [
+        940,
+        520
+      ],
+      "size": [
+        440,
+        440
+      ],
+      "flags": {},
+      "order": 7,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [],
+      "properties": {
+        "Node name for S&R": "Note"
+      },
+      "widgets_values": [
+        "MiniMax H3 Text to Video (WF-01)\n\nBASE: connected presets, 50 steps / video flow 12 / audio flow 3.\n\nTURBO: set Remote LoRA local_path to the SERVER path of minimax_h3_fl2v_turbo_4step_v1.0_768p_bf16.safetensors from lightx2v/Minimax-h3-Turbo. Keep name h3-turbo-v1.0-768p, scale 1.0 and int_id 0. Connect Turbo sampling, Turbo H3 params, and LoRA to Generate Video, replacing the two Base links. Use 5 steps (4 forwards), video flow 6 and audio flow 3. Do not use the ComfyUI-layout artifact.\n\nKeep frame, references and fast_h3 disconnected. 1344x768, 24 FPS, duration 5.167 seconds = 124 frames (17*7+5). The client selects the named 16:9 preset from the dimensions. Save Video preserves video and audio.\n\nSet URL and served model name for your server. See docs/minimax-h3-t2v.md for setup and validation. Source: Comfy-Org/workflow_templates/templates/video_minimax_h3_t2v.json."
+      ],
+      "title": "Setup / Base and Turbo"
+    }
+  ],
+  "links": [
+    [
+      1,
+      1,
+      0,
+      3,
+      2,
+      "SAMPLING_PARAMS"
+    ],
+    [
+      2,
+      2,
+      0,
+      3,
+      4,
+      "VIDEO_PARAMS"
+    ],
+    [
+      3,
+      3,
+      0,
+      4,
+      0,
+      "VIDEO"
+    ]
+  ],
+  "groups": [],
+  "config": {},
+  "extra": {
+    "ds": {
+      "scale": 0.8,
+      "offset": [
+        50,
+        50
+      ]
+    }
+  },
+  "version": 0.4
+}

apps/ComfyUI-vLLM-Omni/example_workflows/vLLM-Omni FastH3 Text to Video.json

@@ -0,0 +1,322 @@
+{
+  "id": "c6db00da-2ccb-4ff8-8334-d67e1a267781",
+  "revision": 0,
+  "last_node_id": 5,
+  "last_link_id": 3,
+  "nodes": [
+    {
+      "id": 1,
+      "type": "MarkdownNote",
+      "pos": [40, 20],
+      "size": [420, 250],
+      "flags": {},
+      "order": 0,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [],
+      "title": "FastH3 server prerequisite",
+      "properties": {},
+      "widgets_values": [
+        "FastH3 must be fused when the vLLM-Omni server starts. Use --task-type fl2va and --lora-path pointing to exactly one FastH3 adapter file. Do not use CPU or layerwise offload. For VSA / Data-Free, also install fastvideo-kernel and use --diffusion-attention-backend FASTVIDEO_VSA --fastvideo-vsa-topk 64."
+      ],
+      "color": "#432",
+      "bgcolor": "#000"
+    },
+    {
+      "id": 2,
+      "type": "VLLMOmniFastH3Deployment",
+      "pos": [40, 320],
+      "size": [420, 160],
+      "flags": {},
+      "order": 1,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "url",
+          "name": "url",
+          "type": "STRING",
+          "widget": {"name": "url"},
+          "link": null
+        },
+        {
+          "localized_name": "model",
+          "name": "model",
+          "type": "STRING",
+          "widget": {"name": "model"},
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "deployment",
+          "name": "deployment",
+          "type": "FASTH3_DEPLOYMENT",
+          "links": [1]
+        }
+      ],
+      "properties": {"Node name for S&R": "VLLMOmniFastH3Deployment"},
+      "widgets_values": [
+        "http://localhost:8000/v1",
+        "MiniMaxAI/MiniMax-H3"
+      ]
+    },
+    {
+      "id": 3,
+      "type": "VLLMOmniDiffusionSampling",
+      "pos": [40, 540],
+      "size": [420, 230],
+      "flags": {},
+      "order": 2,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "n",
+          "name": "n",
+          "type": "INT",
+          "widget": {"name": "n"},
+          "link": null
+        },
+        {
+          "localized_name": "num_inference_steps",
+          "name": "num_inference_steps",
+          "type": "INT",
+          "widget": {"name": "num_inference_steps"},
+          "link": null
+        },
+        {
+          "localized_name": "guidance_scale",
+          "name": "guidance_scale",
+          "type": "FLOAT",
+          "widget": {"name": "guidance_scale"},
+          "link": null
+        },
+        {
+          "localized_name": "true_cfg_scale",
+          "name": "true_cfg_scale",
+          "type": "FLOAT",
+          "widget": {"name": "true_cfg_scale"},
+          "link": null
+        },
+        {
+          "localized_name": "vae_use_slicing",
+          "name": "vae_use_slicing",
+          "type": "BOOLEAN",
+          "widget": {"name": "vae_use_slicing"},
+          "link": null
+        },
+        {
+          "localized_name": "vae_use_tiling",
+          "name": "vae_use_tiling",
+          "type": "BOOLEAN",
+          "widget": {"name": "vae_use_tiling"},
+          "link": null
+        },
+        {
+          "localized_name": "seed",
+          "name": "seed",
+          "type": "INT",
+          "widget": {"name": "seed"},
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "diffusion sampling params",
+          "name": "diffusion sampling params",
+          "type": "SAMPLING_PARAMS",
+          "links": [2]
+        }
+      ],
+      "properties": {"Node name for S&R": "VLLMOmniDiffusionSampling"},
+      "widgets_values": [1, 4, 1, 1, false, false, 1101, "fixed"]
+    },
+    {
+      "id": 4,
+      "type": "VLLMOmniGenerateVideo",
+      "pos": [560, 210],
+      "size": [540, 610],
+      "flags": {},
+      "order": 3,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "frame",
+          "name": "frame",
+          "shape": 7,
+          "type": "IMAGE",
+          "link": null
+        },
+        {
+          "localized_name": "references",
+          "name": "references",
+          "shape": 7,
+          "type": "VIDEO_REFERENCES",
+          "link": null
+        },
+        {
+          "localized_name": "sampling_params",
+          "name": "sampling_params",
+          "shape": 7,
+          "type": "SAMPLING_PARAMS",
+          "link": 2
+        },
+        {
+          "localized_name": "lora",
+          "name": "lora",
+          "shape": 7,
+          "type": "REMOTE_LORA",
+          "link": null
+        },
+        {
+          "localized_name": "model_params",
+          "name": "model_params",
+          "shape": 7,
+          "type": "VIDEO_PARAMS",
+          "link": null
+        },
+        {
+          "localized_name": "fast_h3",
+          "name": "fast_h3",
+          "shape": 7,
+          "type": "FASTH3_DEPLOYMENT",
+          "link": 1
+        },
+        {
+          "localized_name": "url",
+          "name": "url",
+          "type": "STRING",
+          "widget": {"name": "url"},
+          "link": null
+        },
+        {
+          "localized_name": "model",
+          "name": "model",
+          "type": "STRING",
+          "widget": {"name": "model"},
+          "link": null
+        },
+        {
+          "localized_name": "prompt",
+          "name": "prompt",
+          "type": "STRING",
+          "widget": {"name": "prompt"},
+          "link": null
+        },
+        {
+          "localized_name": "negative_prompt",
+          "name": "negative_prompt",
+          "type": "STRING",
+          "widget": {"name": "negative_prompt"},
+          "link": null
+        },
+        {
+          "localized_name": "width",
+          "name": "width",
+          "type": "INT",
+          "widget": {"name": "width"},
+          "link": null
+        },
+        {
+          "localized_name": "height",
+          "name": "height",
+          "type": "INT",
+          "widget": {"name": "height"},
+          "link": null
+        },
+        {
+          "localized_name": "fps",
+          "name": "fps",
+          "type": "INT",
+          "widget": {"name": "fps"},
+          "link": null
+        },
+        {
+          "localized_name": "duration",
+          "name": "duration",
+          "type": "FLOAT",
+          "widget": {"name": "duration"},
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "video",
+          "name": "video",
+          "type": "VIDEO",
+          "links": [3]
+        }
+      ],
+      "properties": {"Node name for S&R": "VLLMOmniGenerateVideo"},
+      "widgets_values": [
+        "http://localhost:8000/v1",
+        "MiniMaxAI/MiniMax-H3",
+        "A cinematic live-action singer performs naturally on a neon-lit stage, with synchronized vocals and environmental sound.",
+        "",
+        1344,
+        768,
+        24,
+        4.458
+      ]
+    },
+    {
+      "id": 5,
+      "type": "SaveVideo",
+      "pos": [1200, 280],
+      "size": [380, 310],
+      "flags": {},
+      "order": 4,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "video",
+          "name": "video",
+          "type": "VIDEO",
+          "link": 3
+        },
+        {
+          "localized_name": "filename_prefix",
+          "name": "filename_prefix",
+          "type": "STRING",
+          "widget": {"name": "filename_prefix"},
+          "link": null
+        },
+        {
+          "localized_name": "format",
+          "name": "format",
+          "type": "COMBO",
+          "widget": {"name": "format"},
+          "link": null
+        },
+        {
+          "localized_name": "codec",
+          "name": "codec",
+          "type": "COMFY_DYNAMICCOMBO_V3",
+          "widget": {"name": "codec"},
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "video",
+          "name": "video",
+          "type": "VIDEO",
+          "links": null
+        }
+      ],
+      "properties": {},
+      "widgets_values": ["video/ComfyUI-FastH3", "auto", "auto"]
+    }
+  ],
+  "links": [
+    [1, 2, 0, 4, 5, "FASTH3_DEPLOYMENT"],
+    [2, 3, 0, 4, 2, "SAMPLING_PARAMS"],
+    [3, 4, 0, 5, 0, "VIDEO"]
+  ],
+  "groups": [],
+  "config": {},
+  "extra": {
+    "workflowRendererVersion": "LG",
+    "ds": {"scale": 0.9, "offset": [100, 100]}
+  },
+  "version": 0.4
+}

apps/ComfyUI-vLLM-Omni/example_workflows/vLLM-Omni MiniMax H3 Image to Video.json

@@ -0,0 +1,219 @@
+{
+  "id": "6676f19a-7f41-4c80-8cb7-f80adcb4ac44",
+  "revision": 0,
+  "last_node_id": 9,
+  "last_link_id": 4,
+  "nodes": [
+    {
+      "id": 1,
+      "type": "LoadImage",
+      "pos": [40, 20],
+      "size": [270, 314],
+      "flags": {},
+      "order": 0,
+      "mode": 0,
+      "inputs": [
+        {"localized_name": "image", "name": "image", "type": "COMBO", "widget": {"name": "image"}, "link": null},
+        {"localized_name": "choose file to upload", "name": "upload", "type": "IMAGEUPLOAD", "widget": {"name": "upload"}, "link": null}
+      ],
+      "outputs": [
+        {"localized_name": "IMAGE", "name": "IMAGE", "type": "IMAGE", "links": [1]},
+        {"localized_name": "MASK", "name": "MASK", "type": "MASK", "links": null}
+      ],
+      "title": "First Frame",
+      "properties": {"Node name for S&R": "LoadImage"},
+      "widgets_values": ["first_frame.png", "image"]
+    },
+    {
+      "id": 2,
+      "type": "LoadImage",
+      "pos": [40, 390],
+      "size": [270, 314],
+      "flags": {},
+      "order": 1,
+      "mode": 0,
+      "inputs": [
+        {"localized_name": "image", "name": "image", "type": "COMBO", "widget": {"name": "image"}, "link": null},
+        {"localized_name": "choose file to upload", "name": "upload", "type": "IMAGEUPLOAD", "widget": {"name": "upload"}, "link": null}
+      ],
+      "outputs": [
+        {"localized_name": "IMAGE", "name": "IMAGE", "type": "IMAGE", "links": null},
+        {"localized_name": "MASK", "name": "MASK", "type": "MASK", "links": null}
+      ],
+      "title": "Last Frame (optional)",
+      "properties": {"Node name for S&R": "LoadImage"},
+      "widgets_values": ["last_frame.png", "image"]
+    },
+    {
+      "id": 3,
+      "type": "MarkdownNote",
+      "pos": [350, 20],
+      "size": [390, 240],
+      "flags": {},
+      "order": 2,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [],
+      "title": "MiniMax H3 I2V modes",
+      "properties": {},
+      "widgets_values": ["This workflow uses the remote vLLM-Omni Generate Video node.\n\n- First-frame only (default): keep First Frame connected.\n- Last-frame only: disconnect First Frame and connect Last Frame to last_frame.\n- First+last-frame: connect both image loaders.\n\nLeave the legacy frame and references inputs disconnected. Select images from ComfyUI's input folder. H3 runs at 24 FPS; the 5.167-second duration resolves to 124 frames, satisfying 17k+5."],
+      "color": "#432",
+      "bgcolor": "#000"
+    },
+    {
+      "id": 4,
+      "type": "VLLMOmniDiffusionSampling",
+      "pos": [350, 300],
+      "size": [300, 226],
+      "flags": {},
+      "order": 3,
+      "mode": 0,
+      "inputs": [
+        {"localized_name": "n", "name": "n", "type": "INT", "widget": {"name": "n"}, "link": null},
+        {"localized_name": "num_inference_steps", "name": "num_inference_steps", "type": "INT", "widget": {"name": "num_inference_steps"}, "link": null},
+        {"localized_name": "guidance_scale", "name": "guidance_scale", "type": "FLOAT", "widget": {"name": "guidance_scale"}, "link": null},
+        {"localized_name": "true_cfg_scale", "name": "true_cfg_scale", "type": "FLOAT", "widget": {"name": "true_cfg_scale"}, "link": null},
+        {"localized_name": "vae_use_slicing", "name": "vae_use_slicing", "type": "BOOLEAN", "widget": {"name": "vae_use_slicing"}, "link": null},
+        {"localized_name": "vae_use_tiling", "name": "vae_use_tiling", "type": "BOOLEAN", "widget": {"name": "vae_use_tiling"}, "link": null},
+        {"localized_name": "seed", "name": "seed", "type": "INT", "widget": {"name": "seed"}, "link": null}
+      ],
+      "outputs": [
+        {"localized_name": "diffusion sampling params", "name": "diffusion sampling params", "type": "SAMPLING_PARAMS", "links": [2]}
+      ],
+      "title": "Base H3 Sampling (50 steps)",
+      "properties": {"Node name for S&R": "VLLMOmniDiffusionSampling"},
+      "widgets_values": [1, 50, 1, 1, false, false, 757358688076805, "fixed"]
+    },
+    {
+      "id": 5,
+      "type": "VLLMOmniMiniMaxH3Params",
+      "pos": [350, 570],
+      "size": [300, 82],
+      "flags": {},
+      "order": 4,
+      "mode": 0,
+      "inputs": [
+        {"localized_name": "audio_flow_shift", "name": "audio_flow_shift", "type": "FLOAT", "widget": {"name": "audio_flow_shift"}, "link": null},
+        {"localized_name": "flow_shift", "name": "flow_shift", "type": "FLOAT", "widget": {"name": "flow_shift"}, "link": null}
+      ],
+      "outputs": [
+        {"localized_name": "MiniMax-H3 video params", "name": "MiniMax-H3 video params", "type": "VIDEO_PARAMS", "links": [3]}
+      ],
+      "title": "Base H3 Flow Shifts",
+      "properties": {"Node name for S&R": "VLLMOmniMiniMaxH3Params"},
+      "widgets_values": [3, 12]
+    },
+    {
+      "id": 6,
+      "type": "VLLMOmniRemoteLoRA",
+      "pos": [700, 550],
+      "size": [360, 150],
+      "flags": {},
+      "order": 5,
+      "mode": 0,
+      "inputs": [
+        {"localized_name": "local_path", "name": "local_path", "type": "STRING", "widget": {"name": "local_path"}, "link": null},
+        {"localized_name": "name", "name": "name", "type": "STRING", "widget": {"name": "name"}, "link": null},
+        {"localized_name": "scale", "name": "scale", "type": "FLOAT", "widget": {"name": "scale"}, "link": null},
+        {"localized_name": "int_id", "name": "int_id", "type": "INT", "widget": {"name": "int_id"}, "link": null}
+      ],
+      "outputs": [
+        {"localized_name": "lora", "name": "lora", "type": "REMOTE_LORA", "links": null}
+      ],
+      "title": "Optional H3 Turbo Remote LoRA",
+      "properties": {"Node name for S&R": "VLLMOmniRemoteLoRA"},
+      "widgets_values": ["minimax_h3_fl2v_turbo_8step_v1.0_768p_bf16.safetensors", "h3-turbo-8step", 1, 0]
+    },
+    {
+      "id": 7,
+      "type": "MarkdownNote",
+      "pos": [700, 300],
+      "size": [420, 210],
+      "flags": {},
+      "order": 6,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [],
+      "title": "Optional Turbo settings",
+      "properties": {},
+      "widgets_values": ["Base H3 is configured by default: 50 steps, flow_shift 12, audio_flow_shift 3.\n\nFor the shown 8-step Diffusers Turbo artifact, set its full local_path on the server, start the server with the matching preloaded LoRA, connect this Remote LoRA to Generate Video, set num_inference_steps to 9, and set flow_shift to 6. Keep audio_flow_shift at 3 and scale at 1.\n\nDo not use the _comfyui_ repack or FastH3 here."],
+      "color": "#432",
+      "bgcolor": "#000"
+    },
+    {
+      "id": 8,
+      "type": "VLLMOmniGenerateVideo",
+      "pos": [1160, 80],
+      "size": [440, 610],
+      "flags": {},
+      "order": 7,
+      "mode": 0,
+      "inputs": [
+        {"localized_name": "frame", "name": "frame", "shape": 7, "type": "IMAGE", "link": null},
+        {"localized_name": "first_frame", "name": "first_frame", "shape": 7, "type": "IMAGE", "link": 1},
+        {"localized_name": "last_frame", "name": "last_frame", "shape": 7, "type": "IMAGE", "link": null},
+        {"localized_name": "references", "name": "references", "shape": 7, "type": "VIDEO_REFERENCES", "link": null},
+        {"localized_name": "sampling_params", "name": "sampling_params", "shape": 7, "type": "SAMPLING_PARAMS", "link": 2},
+        {"localized_name": "lora", "name": "lora", "shape": 7, "type": "REMOTE_LORA", "link": null},
+        {"localized_name": "model_params", "name": "model_params", "shape": 7, "type": "VIDEO_PARAMS", "link": 3},
+        {"localized_name": "fast_h3", "name": "fast_h3", "shape": 7, "type": "FASTH3_DEPLOYMENT", "link": null},
+        {"localized_name": "url", "name": "url", "type": "STRING", "widget": {"name": "url"}, "link": null},
+        {"localized_name": "model", "name": "model", "type": "STRING", "widget": {"name": "model"}, "link": null},
+        {"localized_name": "prompt", "name": "prompt", "type": "STRING", "widget": {"name": "prompt"}, "link": null},
+        {"localized_name": "negative_prompt", "name": "negative_prompt", "type": "STRING", "widget": {"name": "negative_prompt"}, "link": null},
+        {"localized_name": "width", "name": "width", "type": "INT", "widget": {"name": "width"}, "link": null},
+        {"localized_name": "height", "name": "height", "type": "INT", "widget": {"name": "height"}, "link": null},
+        {"localized_name": "fps", "name": "fps", "type": "INT", "widget": {"name": "fps"}, "link": null},
+        {"localized_name": "duration", "name": "duration", "type": "FLOAT", "widget": {"name": "duration"}, "link": null}
+      ],
+      "outputs": [
+        {"localized_name": "video", "name": "video", "type": "VIDEO", "links": [4]}
+      ],
+      "properties": {"Node name for S&R": "VLLMOmniGenerateVideo"},
+      "widgets_values": [
+        "http://localhost:8000/v1",
+        "MiniMaxAI/MiniMax-H3",
+        "A cinematic product shot begins exactly from the supplied keyframe. The subject moves naturally through a dark studio while blue and amber rim lights shift across reflective surfaces. The camera makes a slow controlled orbit. Audio: subtle room tone, precise mechanical details, and a restrained electronic score.",
+        "",
+        1344,
+        768,
+        24,
+        5.167
+      ]
+    },
+    {
+      "id": 9,
+      "type": "SaveVideo",
+      "pos": [1650, 170],
+      "size": [374, 314],
+      "flags": {},
+      "order": 8,
+      "mode": 0,
+      "inputs": [
+        {"localized_name": "video", "name": "video", "type": "VIDEO", "link": 4},
+        {"localized_name": "filename_prefix", "name": "filename_prefix", "type": "STRING", "widget": {"name": "filename_prefix"}, "link": null},
+        {"localized_name": "format", "name": "format", "type": "COMBO", "widget": {"name": "format"}, "link": null},
+        {"localized_name": "codec", "name": "codec", "type": "COMFY_DYNAMICCOMBO_V3", "widget": {"name": "codec"}, "link": null}
+      ],
+      "outputs": [
+        {"localized_name": "video", "name": "video", "type": "VIDEO", "links": null}
+      ],
+      "properties": {},
+      "widgets_values": ["video/MiniMax_H3_I2V", "auto", "auto"]
+    }
+  ],
+  "links": [
+    [1, 1, 0, 8, 1, "IMAGE"],
+    [2, 4, 0, 8, 4, "SAMPLING_PARAMS"],
+    [3, 5, 0, 8, 6, "VIDEO_PARAMS"],
+    [4, 8, 0, 9, 0, "VIDEO"]
+  ],
+  "groups": [
+    {"id": 1, "title": "Portable first/last-frame inputs", "bounding": [10, -30, 760, 770], "color": "#3f789e", "flags": {}},
+    {"id": 2, "title": "H3 sampling and optional Turbo", "bounding": [320, 270, 830, 460], "color": "#8a6d3b", "flags": {}},
+    {"id": 3, "title": "Remote generation and audio-preserving save", "bounding": [1130, 30, 930, 700], "color": "#3f789e", "flags": {}}
+  ],
+  "config": {},
+  "extra": {"workflowRendererVersion": "LG", "ds": {"scale": 0.75, "offset": [180, 120]}},
+  "version": 0.4
+}

apps/ComfyUI-vLLM-Omni/example_workflows/vLLM-Omni MiniMax H3 Video Upscale.json

@@ -0,0 +1,724 @@
+{
+  "version": 0.4,
+  "last_node_id": 12,
+  "last_link_id": 13,
+  "nodes": [
+    {
+      "id": 1,
+      "type": "VLLMOmniGenerateVideo",
+      "pos": [
+        360,
+        20
+      ],
+      "size": [
+        330,
+        480
+      ],
+      "flags": {},
+      "order": 1,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "frame",
+          "type": "IMAGE",
+          "shape": 7,
+          "link": null
+        },
+        {
+          "name": "references",
+          "type": "VIDEO_REFERENCES",
+          "shape": 7,
+          "link": null
+        },
+        {
+          "name": "sampling_params",
+          "type": "SAMPLING_PARAMS",
+          "shape": 7,
+          "link": 11
+        },
+        {
+          "name": "lora",
+          "type": "REMOTE_LORA",
+          "shape": 7,
+          "link": 12
+        },
+        {
+          "name": "model_params",
+          "type": "VIDEO_PARAMS",
+          "shape": 7,
+          "link": 10
+        },
+        {
+          "name": "fast_h3",
+          "type": "FASTH3_DEPLOYMENT",
+          "shape": 7,
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "name": "video",
+          "type": "VIDEO",
+          "links": [
+            1,
+            9
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniGenerateVideo"
+      },
+      "widgets_values": [
+        "http://127.0.0.1:8091/v1",
+        "MiniMaxAI/MiniMax-H3",
+        "A small brass bell hanging from a wooden beam in a quiet garden. A gentle breeze swings the bell and it rings clearly. Static camera, natural daylight, continuous motion, synchronized ringing and soft birdsong.",
+        "",
+        1344,
+        768,
+        24,
+        5.167
+      ]
+    },
+    {
+      "id": 2,
+      "type": "GetVideoComponents",
+      "pos": [
+        730,
+        20
+      ],
+      "size": [
+        250,
+        125
+      ],
+      "flags": {},
+      "order": 1,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "video",
+          "type": "VIDEO",
+          "link": 1
+        }
+      ],
+      "outputs": [
+        {
+          "name": "images",
+          "type": "IMAGE",
+          "links": [
+            2
+          ]
+        },
+        {
+          "name": "audio",
+          "type": "AUDIO",
+          "links": [
+            6
+          ]
+        },
+        {
+          "name": "fps",
+          "type": "FLOAT",
+          "links": [
+            5
+          ]
+        },
+        {
+          "name": "bit_depth",
+          "type": "INT",
+          "links": [
+            7
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "GetVideoComponents"
+      }
+    },
+    {
+      "id": 3,
+      "type": "SeedVR2LoadDiTModel",
+      "pos": [
+        730,
+        290
+      ],
+      "size": [
+        340,
+        245
+      ],
+      "flags": {},
+      "order": 4,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "torch_compile_args",
+          "name": "torch_compile_args",
+          "shape": 7,
+          "type": "TORCH_COMPILE_ARGS",
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "SEEDVR2_DIT",
+          "name": "SEEDVR2_DIT",
+          "type": "SEEDVR2_DIT",
+          "links": [
+            3
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "SeedVR2LoadDiTModel"
+      },
+      "widgets_values": [
+        "seedvr2_ema_3b_fp16.safetensors",
+        "cuda:0",
+        32,
+        false,
+        "cpu",
+        false,
+        "sdpa"
+      ]
+    },
+    {
+      "id": 4,
+      "type": "SeedVR2VideoUpscaler",
+      "pos": [
+        1120,
+        290
+      ],
+      "size": [
+        340,
+        415
+      ],
+      "flags": {},
+      "order": 8,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "image",
+          "name": "image",
+          "type": "IMAGE",
+          "link": 2
+        },
+        {
+          "localized_name": "dit",
+          "name": "dit",
+          "type": "SEEDVR2_DIT",
+          "link": 3
+        },
+        {
+          "localized_name": "vae",
+          "name": "vae",
+          "type": "SEEDVR2_VAE",
+          "link": 13
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "IMAGE",
+          "name": "IMAGE",
+          "type": "IMAGE",
+          "links": [
+            4
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "SeedVR2VideoUpscaler"
+      },
+      "widgets_values": [
+        42,
+        "fixed",
+        1536,
+        0,
+        5,
+        true,
+        "lab",
+        1,
+        0,
+        0.0,
+        0.0,
+        "cpu",
+        false
+      ]
+    },
+    {
+      "id": 5,
+      "type": "CreateVideo",
+      "pos": [
+        1490,
+        290
+      ],
+      "size": [
+        250,
+        175
+      ],
+      "flags": {},
+      "order": 4,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "images",
+          "type": "IMAGE",
+          "link": 4
+        },
+        {
+          "name": "fps",
+          "type": "FLOAT",
+          "widget": {
+            "name": "fps"
+          },
+          "link": 5
+        },
+        {
+          "name": "audio",
+          "type": "AUDIO",
+          "shape": 7,
+          "link": 6
+        },
+        {
+          "name": "bit_depth",
+          "type": "INT",
+          "widget": {
+            "name": "bit_depth"
+          },
+          "link": 7
+        }
+      ],
+      "outputs": [
+        {
+          "name": "VIDEO",
+          "type": "VIDEO",
+          "links": [
+            8
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "CreateVideo"
+      },
+      "widgets_values": [
+        24,
+        8
+      ]
+    },
+    {
+      "id": 6,
+      "type": "SaveVideo",
+      "pos": [
+        1790,
+        290
+      ],
+      "size": [
+        330,
+        280
+      ],
+      "flags": {},
+      "order": 5,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "video",
+          "type": "VIDEO",
+          "link": 8
+        },
+        {
+          "name": "filename_prefix",
+          "type": "STRING",
+          "widget": {
+            "name": "filename_prefix"
+          },
+          "link": null
+        },
+        {
+          "name": "format",
+          "type": "COMBO",
+          "widget": {
+            "name": "format"
+          },
+          "link": null
+        },
+        {
+          "name": "codec",
+          "type": "COMFY_DYNAMICCOMBO_V3",
+          "widget": {
+            "name": "codec"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "name": "video",
+          "type": "VIDEO",
+          "links": []
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "SaveVideo"
+      },
+      "widgets_values": [
+        "WF07/upscaled",
+        "mp4",
+        "auto"
+      ]
+    },
+    {
+      "id": 7,
+      "type": "VLLMOmniMiniMaxH3Params",
+      "pos": [
+        0,
+        20
+      ],
+      "size": [
+        320,
+        120
+      ],
+      "flags": {},
+      "order": 7,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "MiniMax-H3 video params",
+          "type": "VIDEO_PARAMS",
+          "links": [
+            10
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniMiniMaxH3Params"
+      },
+      "widgets_values": [
+        3.0,
+        6.0
+      ]
+    },
+    {
+      "id": 8,
+      "type": "VLLMOmniDiffusionSampling",
+      "pos": [
+        0,
+        200
+      ],
+      "size": [
+        320,
+        280
+      ],
+      "flags": {},
+      "order": 8,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "diffusion sampling params",
+          "type": "SAMPLING_PARAMS",
+          "links": [
+            11
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniDiffusionSampling"
+      },
+      "widgets_values": [
+        1,
+        5,
+        1.0,
+        1.0,
+        false,
+        true,
+        42,
+        "fixed"
+      ]
+    },
+    {
+      "id": 9,
+      "type": "VLLMOmniRemoteLoRA",
+      "pos": [
+        0,
+        530
+      ],
+      "size": [
+        320,
+        210
+      ],
+      "flags": {},
+      "order": 9,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "lora",
+          "type": "REMOTE_LORA",
+          "links": [
+            12
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniRemoteLoRA"
+      },
+      "widgets_values": [
+        "models/minimax-h3-turbo/minimax_h3_fl2v_turbo_4step_v1.0_768p_bf16.safetensors",
+        "h3-turbo-v1.0",
+        1.0,
+        0
+      ]
+    },
+    {
+      "id": 10,
+      "type": "SaveVideo",
+      "pos": [
+        1120,
+        20
+      ],
+      "size": [
+        330,
+        190
+      ],
+      "flags": {},
+      "order": 10,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "video",
+          "type": "VIDEO",
+          "link": 9
+        },
+        {
+          "name": "filename_prefix",
+          "type": "STRING",
+          "widget": {
+            "name": "filename_prefix"
+          },
+          "link": null
+        },
+        {
+          "name": "format",
+          "type": "COMBO",
+          "widget": {
+            "name": "format"
+          },
+          "link": null
+        },
+        {
+          "name": "codec",
+          "type": "COMFY_DYNAMICCOMBO_V3",
+          "widget": {
+            "name": "codec"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "name": "video",
+          "type": "VIDEO",
+          "links": []
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "SaveVideo"
+      },
+      "widgets_values": [
+        "WF07/generated",
+        "mp4",
+        "auto"
+      ],
+      "title": "Save generated video before upscale"
+    },
+    {
+      "id": 11,
+      "type": "Note",
+      "pos": [
+        360,
+        560
+      ],
+      "size": [
+        330,
+        260
+      ],
+      "flags": {},
+      "order": 11,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [],
+      "properties": {
+        "Node name for S&R": "Note"
+      },
+      "widgets_values": [
+        "WF-07: H3 + SeedVR2\n\nInstall ComfyUI-SeedVR2_VideoUpscaler. The DiT and VAE loaders use SeedVR2 3B FP16 and ema_vae_fp16.\n\nSet the H3 service URL/model and the server-side Turbo LoRA path.\n\n1536 is the target short edge for this 1344 x 768 preset (2x). Original audio and FPS connect directly to Create Video.\n\nFor base H3: disconnect LoRA, use 50 steps and video flow shift 12."
+      ]
+    },
+    {
+      "id": 12,
+      "type": "SeedVR2LoadVAEModel",
+      "pos": [
+        730,
+        590
+      ],
+      "size": [
+        340,
+        315
+      ],
+      "flags": {},
+      "order": 5,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "torch_compile_args",
+          "name": "torch_compile_args",
+          "shape": 7,
+          "type": "TORCH_COMPILE_ARGS",
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "SEEDVR2_VAE",
+          "name": "SEEDVR2_VAE",
+          "type": "SEEDVR2_VAE",
+          "links": [
+            13
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "SeedVR2LoadVAEModel"
+      },
+      "widgets_values": [
+        "ema_vae_fp16.safetensors",
+        "cuda:0",
+        true,
+        1024,
+        128,
+        true,
+        768,
+        128,
+        "false",
+        "cpu",
+        false
+      ]
+    }
+  ],
+  "links": [
+    [
+      1,
+      1,
+      0,
+      2,
+      0,
+      "VIDEO"
+    ],
+    [
+      2,
+      2,
+      0,
+      4,
+      0,
+      "IMAGE"
+    ],
+    [
+      3,
+      3,
+      0,
+      4,
+      1,
+      "SEEDVR2_DIT"
+    ],
+    [
+      4,
+      4,
+      0,
+      5,
+      0,
+      "IMAGE"
+    ],
+    [
+      5,
+      2,
+      2,
+      5,
+      1,
+      "FLOAT"
+    ],
+    [
+      6,
+      2,
+      1,
+      5,
+      2,
+      "AUDIO"
+    ],
+    [
+      7,
+      2,
+      3,
+      5,
+      3,
+      "INT"
+    ],
+    [
+      8,
+      5,
+      0,
+      6,
+      0,
+      "VIDEO"
+    ],
+    [
+      9,
+      1,
+      0,
+      10,
+      0,
+      "VIDEO"
+    ],
+    [
+      10,
+      7,
+      0,
+      1,
+      4,
+      "VIDEO_PARAMS"
+    ],
+    [
+      11,
+      8,
+      0,
+      1,
+      2,
+      "SAMPLING_PARAMS"
+    ],
+    [
+      12,
+      9,
+      0,
+      1,
+      3,
+      "REMOTE_LORA"
+    ],
+    [
+      13,
+      12,
+      0,
+      4,
+      2,
+      "SEEDVR2_VAE"
+    ]
+  ],
+  "groups": [],
+  "config": {},
+  "extra": {
+    "ds": {
+      "scale": 0.52,
+      "offset": [
+        100,
+        100
+      ]
+    }
+  }
+}

apps/ComfyUI-vLLM-Omni/example_workflows/vLLM-Omni MiniMax-H3 Reference to Video.json

@@ -0,0 +1,786 @@
+{
+  "id": "e7a4e606-6510-4c4e-b127-9a8b6bf3ea71",
+  "revision": 0,
+  "last_node_id": 11,
+  "last_link_id": 5,
+  "nodes": [
+    {
+      "id": 1,
+      "type": "LoadImage",
+      "pos": [
+        40,
+        100
+      ],
+      "size": [
+        320,
+        300
+      ],
+      "flags": {},
+      "order": 0,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "image",
+          "name": "image",
+          "type": "COMBO",
+          "widget": {
+            "name": "image"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "choose file to upload",
+          "name": "upload",
+          "type": "IMAGEUPLOAD",
+          "widget": {
+            "name": "upload"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "IMAGE",
+          "name": "IMAGE",
+          "type": "IMAGE",
+          "links": [
+            1
+          ]
+        },
+        {
+          "localized_name": "MASK",
+          "name": "MASK",
+          "type": "MASK",
+          "links": null
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "LoadImage"
+      },
+      "widgets_values": [
+        "h3_reference.png",
+        "image"
+      ],
+      "title": "Reference image \u2014 select a local file"
+    },
+    {
+      "type": "LoadVideo",
+      "size": [
+        320,
+        200
+      ],
+      "flags": {},
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "VIDEO",
+          "type": "VIDEO",
+          "links": null
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "LoadVideo"
+      },
+      "widgets_values": [
+        "h3_reference.mp4"
+      ],
+      "id": 2,
+      "pos": [
+        40,
+        470
+      ],
+      "order": 1,
+      "mode": 0,
+      "title": "Optional reference video"
+    },
+    {
+      "id": 3,
+      "type": "LoadAudio",
+      "pos": [
+        40,
+        1100
+      ],
+      "size": [
+        320,
+        160
+      ],
+      "flags": {},
+      "order": 2,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "audio",
+          "name": "audio",
+          "type": "COMBO",
+          "widget": {
+            "name": "audio"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "audioUI",
+          "name": "audioUI",
+          "type": "AUDIO_UI",
+          "widget": {
+            "name": "audioUI"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "choose file to upload",
+          "name": "upload",
+          "type": "AUDIOUPLOAD",
+          "widget": {
+            "name": "upload"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "AUDIO",
+          "name": "AUDIO",
+          "type": "AUDIO",
+          "links": null
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "LoadAudio"
+      },
+      "widgets_values": [
+        "h3_reference.wav",
+        null,
+        ""
+      ],
+      "title": "Optional reference audio"
+    },
+    {
+      "id": 4,
+      "type": "VLLMOmniVideoReferences",
+      "pos": [
+        420,
+        100
+      ],
+      "size": [
+        300,
+        365
+      ],
+      "flags": {},
+      "order": 3,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "image_1",
+          "type": "IMAGE",
+          "shape": 7,
+          "link": 1
+        },
+        {
+          "name": "image_2",
+          "type": "IMAGE",
+          "shape": 7,
+          "link": null
+        },
+        {
+          "name": "audio_1",
+          "type": "AUDIO",
+          "shape": 7,
+          "link": null
+        },
+        {
+          "name": "audio_2",
+          "type": "AUDIO",
+          "shape": 7,
+          "link": null
+        },
+        {
+          "name": "video_1",
+          "type": "VIDEO",
+          "shape": 7,
+          "link": null
+        },
+        {
+          "name": "video_2",
+          "type": "VIDEO",
+          "shape": 7,
+          "link": null
+        },
+        {
+          "name": "image_3",
+          "type": "IMAGE",
+          "shape": 7,
+          "link": null
+        },
+        {
+          "name": "image_4",
+          "type": "IMAGE",
+          "shape": 7,
+          "link": null
+        },
+        {
+          "name": "image_5",
+          "type": "IMAGE",
+          "shape": 7,
+          "link": null
+        },
+        {
+          "name": "image_6",
+          "type": "IMAGE",
+          "shape": 7,
+          "link": null
+        },
+        {
+          "name": "image_7",
+          "type": "IMAGE",
+          "shape": 7,
+          "link": null
+        },
+        {
+          "name": "image_8",
+          "type": "IMAGE",
+          "shape": 7,
+          "link": null
+        },
+        {
+          "name": "image_9",
+          "type": "IMAGE",
+          "shape": 7,
+          "link": null
+        },
+        {
+          "name": "audio_3",
+          "type": "AUDIO",
+          "shape": 7,
+          "link": null
+        },
+        {
+          "name": "video_3",
+          "type": "VIDEO",
+          "shape": 7,
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "references",
+          "name": "references",
+          "type": "VIDEO_REFERENCES",
+          "links": [
+            2
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniVideoReferences"
+      },
+      "widgets_values": [],
+      "title": "References \u2014 up to 12 total"
+    },
+    {
+      "id": 5,
+      "type": "VLLMOmniDiffusionSampling",
+      "pos": [
+        420,
+        530
+      ],
+      "size": [
+        310,
+        275
+      ],
+      "flags": {},
+      "order": 4,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "n",
+          "name": "n",
+          "type": "INT",
+          "widget": {
+            "name": "n"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "num_inference_steps",
+          "name": "num_inference_steps",
+          "type": "INT",
+          "widget": {
+            "name": "num_inference_steps"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "guidance_scale",
+          "name": "guidance_scale",
+          "type": "FLOAT",
+          "widget": {
+            "name": "guidance_scale"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "true_cfg_scale",
+          "name": "true_cfg_scale",
+          "type": "FLOAT",
+          "widget": {
+            "name": "true_cfg_scale"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "vae_use_slicing",
+          "name": "vae_use_slicing",
+          "type": "BOOLEAN",
+          "widget": {
+            "name": "vae_use_slicing"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "vae_use_tiling",
+          "name": "vae_use_tiling",
+          "type": "BOOLEAN",
+          "widget": {
+            "name": "vae_use_tiling"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "seed",
+          "name": "seed",
+          "type": "INT",
+          "widget": {
+            "name": "seed"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "diffusion sampling params",
+          "name": "diffusion sampling params",
+          "type": "SAMPLING_PARAMS",
+          "links": [
+            3
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniDiffusionSampling"
+      },
+      "widgets_values": [
+        1,
+        50,
+        1.0,
+        1.0,
+        false,
+        false,
+        42,
+        "fixed"
+      ],
+      "title": "Sampling \u2014 base 50 / Turbo 5"
+    },
+    {
+      "id": 6,
+      "type": "VLLMOmniMiniMaxH3Params",
+      "pos": [
+        790,
+        750
+      ],
+      "size": [
+        310,
+        105
+      ],
+      "flags": {},
+      "order": 5,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "audio_flow_shift",
+          "name": "audio_flow_shift",
+          "type": "FLOAT",
+          "widget": {
+            "name": "audio_flow_shift"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "flow_shift",
+          "name": "flow_shift",
+          "type": "FLOAT",
+          "widget": {
+            "name": "flow_shift"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "MiniMax-H3 video params",
+          "name": "MiniMax-H3 video params",
+          "type": "VIDEO_PARAMS",
+          "links": [
+            4
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniMiniMaxH3Params"
+      },
+      "widgets_values": [
+        3.0,
+        12.0
+      ],
+      "title": "H3 video/audio flow shifts"
+    },
+    {
+      "type": "VLLMOmniRemoteLoRA",
+      "size": [
+        310,
+        180
+      ],
+      "flags": {},
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "lora",
+          "type": "REMOTE_LORA",
+          "links": null
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniRemoteLoRA"
+      },
+      "widgets_values": [
+        "",
+        "h3-ref2va-turbo",
+        1.0,
+        0
+      ],
+      "id": 7,
+      "pos": [
+        420,
+        940
+      ],
+      "order": 6,
+      "mode": 0,
+      "title": "Optional Turbo LoRA \u2014 connect to enable"
+    },
+    {
+      "id": 8,
+      "type": "VLLMOmniGenerateVideo",
+      "pos": [
+        800,
+        100
+      ],
+      "size": [
+        470,
+        560
+      ],
+      "flags": {},
+      "order": 7,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "frame",
+          "name": "frame",
+          "shape": 7,
+          "type": "IMAGE",
+          "link": null
+        },
+        {
+          "localized_name": "references",
+          "name": "references",
+          "shape": 7,
+          "type": "VIDEO_REFERENCES",
+          "link": 2
+        },
+        {
+          "localized_name": "sampling_params",
+          "name": "sampling_params",
+          "shape": 7,
+          "type": "SAMPLING_PARAMS",
+          "link": 3
+        },
+        {
+          "localized_name": "lora",
+          "name": "lora",
+          "shape": 7,
+          "type": "REMOTE_LORA",
+          "link": null
+        },
+        {
+          "localized_name": "model_params",
+          "name": "model_params",
+          "shape": 7,
+          "type": "VIDEO_PARAMS",
+          "link": 4
+        },
+        {
+          "localized_name": "url",
+          "name": "url",
+          "type": "STRING",
+          "widget": {
+            "name": "url"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "model",
+          "name": "model",
+          "type": "STRING",
+          "widget": {
+            "name": "model"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "prompt",
+          "name": "prompt",
+          "type": "STRING",
+          "widget": {
+            "name": "prompt"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "negative_prompt",
+          "name": "negative_prompt",
+          "type": "STRING",
+          "widget": {
+            "name": "negative_prompt"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "width",
+          "name": "width",
+          "type": "INT",
+          "widget": {
+            "name": "width"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "height",
+          "name": "height",
+          "type": "INT",
+          "widget": {
+            "name": "height"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "fps",
+          "name": "fps",
+          "type": "INT",
+          "widget": {
+            "name": "fps"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "duration",
+          "name": "duration",
+          "type": "FLOAT",
+          "widget": {
+            "name": "duration"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "video",
+          "name": "video",
+          "type": "VIDEO",
+          "links": [
+            5
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniGenerateVideo"
+      },
+      "widgets_values": [
+        "http://localhost:8000/v1",
+        "MiniMaxAI/MiniMax-H3",
+        "Use the subject and visual style from <Picture 1>. Create a cinematic shot with a gentle camera move and natural ambient sound.",
+        "",
+        1344,
+        768,
+        24,
+        5.167
+      ],
+      "title": "MiniMax-H3 Reference to Video"
+    },
+    {
+      "id": 9,
+      "type": "SaveVideo",
+      "pos": [
+        1360,
+        100
+      ],
+      "size": [
+        410,
+        420
+      ],
+      "flags": {},
+      "order": 8,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "video",
+          "name": "video",
+          "type": "VIDEO",
+          "link": 5
+        },
+        {
+          "localized_name": "filename_prefix",
+          "name": "filename_prefix",
+          "type": "STRING",
+          "widget": {
+            "name": "filename_prefix"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "format",
+          "name": "format",
+          "type": "COMBO",
+          "widget": {
+            "name": "format"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "codec",
+          "name": "codec",
+          "type": "COMFY_DYNAMICCOMBO_V3",
+          "widget": {
+            "name": "codec"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "video",
+          "name": "video",
+          "type": "VIDEO",
+          "links": null
+        }
+      ],
+      "properties": {},
+      "widgets_values": [
+        "video/MiniMax-H3-Ref2VA",
+        "mp4",
+        "h264"
+      ],
+      "title": "Save generated video with audio"
+    },
+    {
+      "id": 10,
+      "type": "MarkdownNote",
+      "pos": [
+        1360,
+        590
+      ],
+      "size": [
+        440,
+        595
+      ],
+      "flags": {},
+      "order": 9,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [],
+      "title": "Reference inputs and output settings",
+      "properties": {},
+      "widgets_values": [
+        "## Reference to Video\nSelect an image, set the server URL and served model name, then run.\n\nConnect the optional video/audio loaders to Video References, or duplicate loaders for more inputs: up to 9 images, 3 videos, 3 audio clips, 12 total. At least one image or video is required.\n\nSlots are numbered separately per media type. Empty slots are skipped. Update prompt tags such as <Picture 1>, <Video 1>, <Audio 1> to match the connected references. Videos keep their own soundtrack in the reference upload.\n\n**Output:** 24 FPS; 124 frames (~5.17 s). Valid frame counts are 107, 124, 141, 158, 175, 192, 209, 226, 243, 260, 277, 294, 311, 328, 345.\n\n**Width x height:** 1344x768 (16:9), 1024x768 (4:3), 768x768 (1:1), 768x1024 (3:4), 768x1344 (9:16).\n\nEach reference video/audio clip must be 2\u201315 s. Total reference-video duration <=15 s; total standalone-audio duration <=15 s.\n\nRequires the expanded Video References node and the audio-preserving video output fix. See the extension README for setup and validation status."
+      ],
+      "color": "#432",
+      "bgcolor": "#000"
+    },
+    {
+      "id": 11,
+      "type": "MarkdownNote",
+      "pos": [
+        790,
+        940
+      ],
+      "size": [
+        480,
+        455
+      ],
+      "flags": {},
+      "order": 10,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [],
+      "title": "Optional Turbo \u2014 same workflow",
+      "properties": {},
+      "widgets_values": [
+        "## Turbo Ref2VA\nBase mode: leave LoRA disconnected; use 50 steps, video flow shift 12, audio flow shift 3.\n\nFor the 4-step v0.1 Ref2VA Turbo adapter:\n1. Start a Ref2VA-only server with the Diffusers adapter loaded (see README).\n2. Set LoRA local_path to its path on the server, then connect LoRA to Generate Video.\n3. Set Sampling num_inference_steps to **5**; keep shifts at **12 / 3**, scale **1**.\n\nvLLM-Omni counts sigma points: 5 points give 4 denoiser evaluations. Use minimax_h3_ref2v_turbo_4step_v0.1_bf16.safetensors, not the _comfyui_ export.\n\nDisconnect LoRA and restore 50 steps to return to base mode. FastH3 is a separate T2VA deployment, not this option.\n\nAdapted from the official ComfyUI MiniMax-H3 R2V workflow: https://github.com/Comfy-Org/workflow_templates/blob/main/templates/video_minimax_h3_r2v.json"
+      ],
+      "color": "#432",
+      "bgcolor": "#000"
+    }
+  ],
+  "links": [
+    [
+      1,
+      1,
+      0,
+      4,
+      0,
+      "IMAGE"
+    ],
+    [
+      2,
+      4,
+      0,
+      8,
+      1,
+      "VIDEO_REFERENCES"
+    ],
+    [
+      3,
+      5,
+      0,
+      8,
+      2,
+      "SAMPLING_PARAMS"
+    ],
+    [
+      4,
+      6,
+      0,
+      8,
+      4,
+      "VIDEO_PARAMS"
+    ],
+    [
+      5,
+      8,
+      0,
+      9,
+      0,
+      "VIDEO"
+    ]
+  ],
+  "groups": [],
+  "config": {},
+  "extra": {
+    "workflowRendererVersion": "LG",
+    "ds": {
+      "scale": 0.85,
+      "offset": [
+        100,
+        100
+      ]
+    }
+  },
+  "version": 0.4
+}

apps/ComfyUI-vLLM-Omni/example_workflows/vLLM-Omni Music Generation.json

@@ -0,0 +1,206 @@
+{
+  "id": "4949c8da-12ff-449a-94b7-948bde9c39dd",
+  "revision": 0,
+  "last_node_id": 3,
+  "last_link_id": 1,
+  "nodes": [
+    {
+      "id": 1,
+      "type": "VLLMOmniGenerateMusic",
+      "pos": [
+        100,
+        100
+      ],
+      "size": [
+        470,
+        491
+      ],
+      "flags": {},
+      "order": 0,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "url",
+          "type": "STRING",
+          "widget": {
+            "name": "url"
+          },
+          "link": null
+        },
+        {
+          "name": "model",
+          "type": "STRING",
+          "widget": {
+            "name": "model"
+          },
+          "link": null
+        },
+        {
+          "name": "instructions",
+          "type": "STRING",
+          "widget": {
+            "name": "instructions"
+          },
+          "link": null
+        },
+        {
+          "name": "lyrics",
+          "type": "STRING",
+          "widget": {
+            "name": "lyrics"
+          },
+          "link": null
+        },
+        {
+          "name": "max_duration_seconds",
+          "type": "FLOAT",
+          "widget": {
+            "name": "max_duration_seconds"
+          },
+          "link": null
+        },
+        {
+          "name": "seed",
+          "type": "INT",
+          "widget": {
+            "name": "seed"
+          },
+          "link": null
+        },
+        {
+          "name": "response_format",
+          "type": "COMBO",
+          "widget": {
+            "name": "response_format"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "name": "audio",
+          "type": "AUDIO",
+          "links": [
+            1
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VLLMOmniGenerateMusic"
+      },
+      "widgets_values": [
+        "http://localhost:8000/v1",
+        "MiniMaxAI/MiniMax-Music3",
+        "Futuristic melodic EDM / progressive house. Reflective verses about memory and digital identity build into an uplifting, hook-driven chorus, with an emotive lead vocal, bright layered synths, pulsing bass, crisp four-on-the-floor drums, subtle glitch textures, and a wide polished festival mix.",
+        "[Verse 1]\nUpload my heart to the cloud tonight\nSave every spark of neon light\nIf this body starts to fade away\nLet my code keep dancing in the data stream\nStay\n[Chorus]\nCloud copy of me\nLiving on when these bones finally break free\nEvery joke, every scar\nEvery cheap memory we made\nSave it, save it\nCloud copy of me\nIf the screen goes dark, I\u2019m still in the machine\nEvery glitch, every dream\nEvery version you ever did see\nSave it, save it\n[Verse 2]\nTag every laugh in a secret folder\nArchive the nights we outran the thunder\nIf the wiring turns to rust and gray\nHit restore\u2026 press replay\n[Chorus]\nCloud copy of me\nLiving on when these bones finally break free\nEvery joke, every scar\nEvery cheap memory we made\nSave it, save it\nCloud copy of me\nIf the screen goes dark, I\u2019m still in the machine\nEvery glitch, every dream\nEvery version you ever did see\nSave it, save it (oh)\n[Bridge \u2013 new, optional but recommended]\nEven the crashes, the late-night confessions\nThe parts of me I never learned to mention\nDon\u2019t let the silence swallow me whole\nKeep a ghost in the glow\u2026\nSo I never go\n[Final Chorus \u2013 softer / layered]\nCloud copy of me\nStill laughing somewhere in the binary\nEvery joke, every scar\nEvery moment you saved for me\nSave it, save it\nCloud copy of me\nWhen the lights go out, I\u2019ll still be in the machine\nDancing in the code that you still believe\nSave it\u2026\nSave me",
+        300.0,
+        0,
+        "fixed",
+        "wav"
+      ]
+    },
+    {
+      "id": 2,
+      "type": "SaveAudioAdvanced",
+      "pos": [
+        650,
+        100
+      ],
+      "size": [
+        320,
+        491
+      ],
+      "flags": {},
+      "order": 1,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "audio",
+          "type": "AUDIO",
+          "link": 1
+        },
+        {
+          "name": "filename_prefix",
+          "type": "STRING",
+          "widget": {
+            "name": "filename_prefix"
+          },
+          "link": null
+        },
+        {
+          "name": "format",
+          "type": "COMFY_DYNAMICCOMBO_V3",
+          "widget": {
+            "name": "format"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "name": "audio",
+          "type": "AUDIO",
+          "links": null
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "SaveAudioAdvanced"
+      },
+      "widgets_values": [
+        "audio/vLLM-Omni-Music",
+        "mp3",
+        "V0"
+      ]
+    },
+    {
+      "id": 3,
+      "type": "MarkdownNote",
+      "title": "Music Generation Guide",
+      "pos": [
+        -507,
+        100
+      ],
+      "size": [
+        527,
+        491
+      ],
+      "flags": {},
+      "order": 2,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [],
+      "properties": {},
+      "widgets_values": [
+        "## MiniMax Music 3\n\nThe **Generate Music** node currently supports **MiniMax Music 3 only**. The parameters below apply to this model.\n\nA music generation model by MiniMax that creates complete songs up to 5 minutes long with stable structure and high audio quality. Two inputs drive it: **Caption** (structured music description: style, mood, vocals, arrangement) and **Lyrics** (with `[intro]` `[verse]` `[chorus]` `[bridge]` `[outro]` tags controlling song structure).\n\n### Parameters\n\n| Parameter | Description |\n|---|---|\n| url | vLLM-Omni service URL including `/v1`. Default: `http://localhost:8000/v1`. |\n| model | Served model name. Default: `MiniMaxAI/MiniMax-Music3`. |\n| caption | Music description, sent as `instructions`. Use Global Metadata \u2192 Vocal Details \u2192 Arrangement to describe style, vocals, and instrumentation. |\n| lyrics | Words to sing with section tags. Put each tag on its own line; tags guide rather than strictly enforce song structure. |\n| max_duration_seconds | Maximum duration, default 300 seconds; range 1\u2013360, step 0.01. Converted to tokens at 25 frames/s, rounded down. Generation may end earlier. |\n| seed | Random seed, default 0; range 0\u20139007199254740991. Keep inputs and runtime settings unchanged for repeatability. |\n| control_after_generate | Seed behavior after generation: fixed, randomize, increment, or decrement. Template default: fixed. |\n| response_format | API audio format, default WAV. Options: WAV, MP3, FLAC, Opus, AAC; AAC is currently unsupported by the vLLM-Omni speech API. |\n\n### Prompt Writing Guide\n\nYou can find an official [Music Caption Rewriter Skill here](https://github.com/MiniMax-AI/MiniMax-Music3)\n"
+      ],
+      "widgets_values_named": {
+        "text": "## MiniMax Music 3\n\nThe **Generate Music** node currently supports **MiniMax Music 3 only**. The parameters below apply to this model.\n\nA music generation model by MiniMax that creates complete songs up to 5 minutes long with stable structure and high audio quality. Two inputs drive it: **Caption** (structured music description: style, mood, vocals, arrangement) and **Lyrics** (with `[intro]` `[verse]` `[chorus]` `[bridge]` `[outro]` tags controlling song structure).\n\n### Parameters\n\n| Parameter | Description |\n|---|---|\n| url | vLLM-Omni service URL including `/v1`. Default: `http://localhost:8000/v1`. |\n| model | Served model name. Default: `MiniMaxAI/MiniMax-Music3`. |\n| caption | Music description, sent as `instructions`. Use Global Metadata \u2192 Vocal Details \u2192 Arrangement to describe style, vocals, and instrumentation. |\n| lyrics | Words to sing with section tags. Put each tag on its own line; tags guide rather than strictly enforce song structure. |\n| max_duration_seconds | Maximum duration, default 300 seconds; range 1\u2013360, step 0.01. Converted to tokens at 25 frames/s, rounded down. Generation may end earlier. |\n| seed | Random seed, default 0; range 0\u20139007199254740991. Keep inputs and runtime settings unchanged for repeatability. |\n| control_after_generate | Seed behavior after generation: fixed, randomize, increment, or decrement. Template default: fixed. |\n| response_format | API audio format, default WAV. Options: WAV, MP3, FLAC, Opus, AAC; AAC is currently unsupported by the vLLM-Omni speech API. |\n\n### Prompt Writing Guide\n\nYou can find an official [Music Caption Rewriter Skill here](https://github.com/MiniMax-AI/MiniMax-Music3)\n"
+      },
+      "color": "#333",
+      "bgcolor": "#535353"
+    }
+  ],
+  "links": [
+    [
+      1,
+      1,
+      0,
+      2,
+      0,
+      "AUDIO"
+    ]
+  ],
+  "groups": [],
+  "config": {},
+  "extra": {
+    "workflowRendererVersion": "LG",
+    "ds": {
+      "scale": 1.147,
+      "offset": [
+        899,
+        259
+      ]
+    }
+  },
+  "version": 0.4
+}

apps/ComfyUI-vLLM-Omni/example_workflows/vLLM-Omni Video Generation.json

@@ -1 +1 @@
-{"id":"3aea1d08-7de5-4997-ba71-b2f66ad72a21","revision":0,"last_node_id":24,"last_link_id":35,"nodes":[{"id":13,"type":"SaveVideo","pos":[1315.729153459605,-137.92038948394355],"size":[374,313.7692307692308],"flags":{},"order":9,"mode":0,"inputs":[{"localized_name":"video","name":"video","type":"VIDEO","link":32},{"localized_name":"filename_prefix","name":"filename_prefix","type":"STRING","widget":{"name":"filename_prefix"},"link":null},{"localized_name":"format","name":"format","type":"COMBO","widget":{"name":"format"},"link":null},{"localized_name":"codec","name":"codec","type":"COMFY_DYNAMICCOMBO_V3","widget":{"name":"codec"},"link":null}],"outputs":[{"localized_name":"video","name":"video","type":"VIDEO","links":null}],"properties":{},"widgets_values":["video/ComfyUI","auto","auto"]},{"id":10,"type":"MarkdownNote","pos":[183.74636797766794,-571.2529158046766],"size":[240.3326478922474,136.3505820791881],"flags":{},"order":0,"mode":0,"inputs":[],"outputs":[],"title":"Note: Task and Input","properties":{},"widgets_values":["vLLM-Omni nodes are categorized based on the output modality. The \"Generate Video\" node supports both *text-to-video*, *frame-to-video*, and *reference-to-video* generation. The node will create corresponding payload depending on whether an input image is present or not."],"color":"#432","bgcolor":"#000"},{"id":4,"type":"LoadImage","pos":[452.07189605801125,-569.7242536425791],"size":[270,314],"flags":{},"order":1,"mode":0,"inputs":[{"localized_name":"image","name":"image","type":"COMBO","widget":{"name":"image"},"link":null},{"localized_name":"choose file to upload","name":"upload","type":"IMAGEUPLOAD","widget":{"name":"upload"},"link":null}],"outputs":[{"localized_name":"IMAGE","name":"IMAGE","type":"IMAGE","links":[29]},{"localized_name":"MASK","name":"MASK","type":"MASK","links":null}],"properties":{"Node name for S&R":"LoadImage"},"widgets_values":["rap.png","image"]},{"id":23,"type":"LoadImage","pos":[9.206110246057657,-7.6955291509170785],"size":[270,314],"flags":{},"order":2,"mode":0,"inputs":[{"localized_name":"image","name":"image","type":"COMBO","widget":{"name":"image"},"link":null},{"localized_name":"choose file to upload","name":"upload","type":"IMAGEUPLOAD","widget":{"name":"upload"},"link":null}],"outputs":[{"localized_name":"IMAGE","name":"IMAGE","type":"IMAGE","links":[34]},{"localized_name":"MASK","name":"MASK","type":"MASK","links":null}],"properties":{"Node name for S&R":"LoadImage"},"widgets_values":["white-cat.jpg","image"]},{"id":22,"type":"MarkdownNote","pos":[-9.783113411416666,-148.8790300038605],"size":[297.3012007349086,88],"flags":{},"order":3,"mode":0,"inputs":[],"outputs":[],"title":"Note: Reference to video (r2v)","properties":{},"widgets_values":["For r2v tasks, connect this Video Reference block to the main Generate Video block. Then disconnect the \"frame\" input above."],"color":"#432","bgcolor":"#000"},{"id":24,"type":"LoadAudio","pos":[317.9929742385805,-141.44821500065095],"size":[282.798828125,136],"flags":{},"order":4,"mode":0,"inputs":[{"localized_name":"audio","name":"audio","type":"COMBO","widget":{"name":"audio"},"link":null},{"localized_name":"audioUI","name":"audioUI","type":"AUDIO_UI","widget":{"name":"audioUI"},"link":null},{"localized_name":"choose file to upload","name":"upload","type":"AUDIOUPLOAD","widget":{"name":"upload"},"link":null}],"outputs":[{"localized_name":"AUDIO","name":"AUDIO","type":"AUDIO","links":[35]}],"properties":{"Node name for S&R":"LoadAudio"},"widgets_values":["nothing-beats-a-jet2-holiday.mp3",null,""]},{"id":21,"type":"VLLMOmniVideoReferences","pos":[376.6129172897595,105.41557277556292],"size":[156.9380859375,126],"flags":{},"order":7,"mode":0,"inputs":[{"localized_name":"image_1","name":"image_1","shape":7,"type":"IMAGE","link":34},{"localized_name":"image_2","name":"image_2","shape":7,"type":"IMAGE","link":null},{"localized_name":"audio_1","name":"audio_1","shape":7,"type":"AUDIO","link":35},{"localized_name":"audio_2","name":"audio_2","shape":7,"type":"AUDIO","link":null},{"localized_name":"video_1","name":"video_1","shape":7,"type":"VIDEO","link":null},{"localized_name":"video_2","name":"video_2","shape":7,"type":"VIDEO","link":null}],"outputs":[{"localized_name":"references","name":"references","type":"VIDEO_REFERENCES","links":null}],"properties":{"Node name for S&R":"VLLMOmniVideoReferences"},"widgets_values":[]},{"id":8,"type":"VLLMOmniDiffusionSampling","pos":[83.24620208113915,422.08543477542605],"size":[284.205078125,226],"flags":{},"order":5,"mode":0,"inputs":[{"localized_name":"n","name":"n","type":"INT","widget":{"name":"n"},"link":null},{"localized_name":"num_inference_steps","name":"num_inference_steps","type":"INT","widget":{"name":"num_inference_steps"},"link":null},{"localized_name":"guidance_scale","name":"guidance_scale","type":"FLOAT","widget":{"name":"guidance_scale"},"link":null},{"localized_name":"true_cfg_scale","name":"true_cfg_scale","type":"FLOAT","widget":{"name":"true_cfg_scLine truncated
+{"id":"3aea1d08-7de5-4997-ba71-b2f66ad72a21","revision":0,"last_node_id":24,"last_link_id":35,"nodes":[{"id":13,"type":"SaveVideo","pos":[1315.729153459605,-137.92038948394355],"size":[374,313.7692307692308],"flags":{},"order":9,"mode":0,"inputs":[{"localized_name":"video","name":"video","type":"VIDEO","link":32},{"localized_name":"filename_prefix","name":"filename_prefix","type":"STRING","widget":{"name":"filename_prefix"},"link":null},{"localized_name":"format","name":"format","type":"COMBO","widget":{"name":"format"},"link":null},{"localized_name":"codec","name":"codec","type":"COMFY_DYNAMICCOMBO_V3","widget":{"name":"codec"},"link":null}],"outputs":[{"localized_name":"video","name":"video","type":"VIDEO","links":null}],"properties":{},"widgets_values":["video/ComfyUI","auto","auto"]},{"id":10,"type":"MarkdownNote","pos":[183.74636797766794,-571.2529158046766],"size":[240.3326478922474,136.3505820791881],"flags":{},"order":0,"mode":0,"inputs":[],"outputs":[],"title":"Note: Task and Input","properties":{},"widgets_values":["vLLM-Omni nodes are categorized based on the output modality. The \"Generate Video\" node supports both *text-to-video*, *frame-to-video*, and *reference-to-video* generation. The node will create corresponding payload depending on whether an input image is present or not."],"color":"#432","bgcolor":"#000"},{"id":4,"type":"LoadImage","pos":[452.07189605801125,-569.7242536425791],"size":[270,314],"flags":{},"order":1,"mode":0,"inputs":[{"localized_name":"image","name":"image","type":"COMBO","widget":{"name":"image"},"link":null},{"localized_name":"choose file to upload","name":"upload","type":"IMAGEUPLOAD","widget":{"name":"upload"},"link":null}],"outputs":[{"localized_name":"IMAGE","name":"IMAGE","type":"IMAGE","links":[29]},{"localized_name":"MASK","name":"MASK","type":"MASK","links":null}],"properties":{"Node name for S&R":"LoadImage"},"widgets_values":["rap.png","image"]},{"id":23,"type":"LoadImage","pos":[9.206110246057657,-7.6955291509170785],"size":[270,314],"flags":{},"order":2,"mode":0,"inputs":[{"localized_name":"image","name":"image","type":"COMBO","widget":{"name":"image"},"link":null},{"localized_name":"choose file to upload","name":"upload","type":"IMAGEUPLOAD","widget":{"name":"upload"},"link":null}],"outputs":[{"localized_name":"IMAGE","name":"IMAGE","type":"IMAGE","links":[34]},{"localized_name":"MASK","name":"MASK","type":"MASK","links":null}],"properties":{"Node name for S&R":"LoadImage"},"widgets_values":["white-cat.jpg","image"]},{"id":22,"type":"MarkdownNote","pos":[-9.783113411416666,-148.8790300038605],"size":[297.3012007349086,88],"flags":{},"order":3,"mode":0,"inputs":[],"outputs":[],"title":"Note: Reference to video (r2v)","properties":{},"widgets_values":["For r2v tasks, connect this Video Reference block to the main Generate Video block. Then disconnect the \"frame\" input above."],"color":"#432","bgcolor":"#000"},{"id":24,"type":"LoadAudio","pos":[317.9929742385805,-141.44821500065095],"size":[282.798828125,136],"flags":{},"order":4,"mode":0,"inputs":[{"localized_name":"audio","name":"audio","type":"COMBO","widget":{"name":"audio"},"link":null},{"localized_name":"audioUI","name":"audioUI","type":"AUDIO_UI","widget":{"name":"audioUI"},"link":null},{"localized_name":"choose file to upload","name":"upload","type":"AUDIOUPLOAD","widget":{"name":"upload"},"link":null}],"outputs":[{"localized_name":"AUDIO","name":"AUDIO","type":"AUDIO","links":[35]}],"properties":{"Node name for S&R":"LoadAudio"},"widgets_values":["nothing-beats-a-jet2-holiday.mp3",null,""]},{"id":21,"type":"VLLMOmniVideoReferences","pos":[376.6129172897595,105.41557277556292],"size":[156.9380859375,126],"flags":{},"order":7,"mode":0,"inputs":[{"localized_name":"image_1","name":"image_1","shape":7,"type":"IMAGE","link":34},{"localized_name":"image_2","name":"image_2","shape":7,"type":"IMAGE","link":null},{"localized_name":"audio_1","name":"audio_1","shape":7,"type":"AUDIO","link":35},{"localized_name":"audio_2","name":"audio_2","shape":7,"type":"AUDIO","link":null},{"localized_name":"video_1","name":"video_1","shape":7,"type":"VIDEO","link":null},{"localized_name":"video_2","name":"video_2","shape":7,"type":"VIDEO","link":null}],"outputs":[{"localized_name":"references","name":"references","type":"VIDEO_REFERENCES","links":null}],"properties":{"Node name for S&R":"VLLMOmniVideoReferences"},"widgets_values":[]},{"id":8,"type":"VLLMOmniDiffusionSampling","pos":[83.24620208113915,422.08543477542605],"size":[284.205078125,226],"flags":{},"order":5,"mode":0,"inputs":[{"localized_name":"n","name":"n","type":"INT","widget":{"name":"n"},"link":null},{"localized_name":"num_inference_steps","name":"num_inference_steps","type":"INT","widget":{"name":"num_inference_steps"},"link":null},{"localized_name":"guidance_scale","name":"guidance_scale","type":"FLOAT","widget":{"name":"guidance_scale"},"link":null},{"localized_name":"true_cfg_scale","name":"true_cfg_scale","type":"FLOAT","widget":{"name":"true_cfg_scLine truncated

apps/ComfyUI-vLLM-Omni/web/main.js

@@ -1,19 +1,154 @@
 /**
- * @file This file is intended to add dynamic fields to vLLM-Omni nodes
- * based on widget (in-node form fields) and input (connection link) values and changes.
- * However, this functionality is currently disabled/commented out.
- * Because it introduces too much complexity,
- * and it may even conflict with the current backend (Python) validation for unknown reasons (pending ComfyUI upstream fixes).
+ * @file Front-end extension for the vLLM-Omni nodes.
+ *
+ * Branding. Every vLLM-Omni node draws the vLLM-Omni "v" mark where LiteGraph
+ * would otherwise draw its default title dot, and is tinted by the kind of value
+ * it produces: blue nodes reach a server, amber and purple ones only assemble
+ * parameters, red ones pin what the server has loaded, teal ones carry media.
+ * A graph is then readable at a glance, and a node that turns out to be the
+ * wrong colour for where it sits is usually wired wrong.
+ *
+ * The tint is keyed off each node's declared output types rather than a list of
+ * node names, so a node added later is themed as soon as it returns one of the
+ * existing types.
+ *
+ * It also migrates Generate Video graphs saved before `duration` replaced the
+ * `num_frames` widget. See migrateNumFramesToDuration below.
+ *
+ * This file also used to host dynamic fields on vLLM-Omni nodes, driven by
+ * widget (in-node form fields) and input (connection link) values and changes.
+ * That is deliberately absent: it introduced too much complexity, and it even
+ * conflicted with the current backend (Python) validation for unknown reasons
+ * (pending ComfyUI upstream fixes).
  */
 
 import { app } from "../../scripts/app.js";
+
+const MARK_URL = new URL("./vllm-omni-mark.svg", import.meta.url).href;
+const MARK_ASPECT = 157 / 165;
+
+/**
+ * Title bar and body tints, one per node family.
+ *
+ * The hues are the logo's: the amber and blue of the "v" mark, and three stops
+ * of the gradient ring in the "O" (#9128A3, #CC5260, and the teal neighbouring
+ * its blue). All of them are dark because LiteGraph paints the title bar darker
+ * than the body and expects both to sit under light text; the ComfyUI light
+ * theme lightens custom node colours on its own.
+ */
+const FAMILIES = {
+    generate: { color: "#152A43", bgcolor: "#1F4269" },
+    sampling: { color: "#3A2C12", bgcolor: "#5C461C" },
+    model: { color: "#2E1233", bgcolor: "#4A1C52" },
+    deployment: { color: "#3A1A1F", bgcolor: "#5C2931" },
+    media: { color: "#0F2E2A", bgcolor: "#194A44" },
+};
+
+/** Output type -> family. Types absent here fall through to "generate". */
+const FAMILY_BY_OUTPUT = {
+    SAMPLING_PARAMS: "sampling",
+    TTS_PARAMS: "model",
+    VIDEO_PARAMS: "model",
+    REMOTE_LORA: "deployment",
+    FASTH3_DEPLOYMENT: "deployment",
+    VIDEO_REFERENCES: "media",
+    LATENT_MASK_EDITING: "media",
+};
+
+function familyOf(nodeData) {
+    for (const type of nodeData.output ?? []) {
+        const family = FAMILY_BY_OUTPUT[type];
+        if (family) {
+            return family;
+        }
+    }
+    // What is left returns a generated artifact (IMAGE, VIDEO, AUDIO, STRING),
+    // which is only ever produced by a node that talks to a vLLM-Omni server.
+    return "generate";
+}
+
+/**
+ * Rewrite a pre-rename Generate Video graph in place.
+ *
+ * `duration` (seconds) took over the slot `num_frames` used to occupy, and
+ * ComfyUI restores widget values positionally, so a graph saved before the
+ * rename reads its frame count as a duration: the workflow this repository
+ * shipped stored 120 frames at 24 fps, which would come back as a 120-second
+ * request. Saved files still name the old widget under `inputs`, and that is
+ * what makes the rewrite both detectable and safe to apply exactly once -- once
+ * rewritten the node serializes `duration`, so a later load finds nothing to do.
+ */
+function migrateNumFramesToDuration(node, info) {
+    const savedBeforeRename = (info?.inputs ?? []).some((input) => input?.widget?.name === "num_frames");
+    if (!savedBeforeRename) {
+        return
+    }
+    const duration = node.widgets?.find((widget) => widget.name === "duration");
+    const fps = node.widgets?.find((widget) => widget.name === "fps");
+    if (!duration || !fps) {
+        return
+    }
+    // configure() has already placed the stale frame count in the duration widget.
+    const frames = Number(duration.value);
+    const rate = Number(fps.value);
+    if (!Number.isFinite(frames) || !Number.isFinite(rate) || rate <= 0) {
+        return
+    }
+    const seconds = Math.max(duration.options?.min ?? 0.1, Math.round((frames / rate) * 1000) / 1000);
+    console.info(
+        `vLLM-Omni: migrated Generate Video "${node.title}" from ${frames} frames to ${seconds}s at ${rate} fps.`,
+    );
+    duration.value = seconds;
+}
+
+const mark = new Image();
+let markReady = false;
+mark.addEventListener("load", () => {
+    markReady = true;
+    // The image resolves after the first frames are already on screen.
+    app.graph?.setDirtyCanvas?.(true, true);
+});
+mark.src = MARK_URL;
+
 app.registerExtension({
     name: "vllm.vllm_omni",
     async beforeRegisterNodeDef(nodeType, nodeData, app) {
         if (!nodeData.name.startsWith("VLLMOmni")) {
             return
         }
-        // Stub frontend plugin for now
+
+        const { color, bgcolor } = FAMILIES[familyOf(nodeData)];
+        // On the prototype, not the instance: a colour the user picks by hand is
+        // written to the instance and shadows this, so manual overrides survive.
+        nodeType.prototype.color = color;
+        nodeType.prototype.bgcolor = bgcolor;
+
+        if (nodeData.name === "VLLMOmniGenerateVideo") {
+            const onConfigure = nodeType.prototype.onConfigure;
+            nodeType.prototype.onConfigure = function (info) {
+                onConfigure?.apply(this, arguments);
+                migrateNumFramesToDuration(this, info);
+            };
+        }
+
+        // Defining this replaces the default dot outright — LiteGraph draws its
+        // own square or circle only for nodes that leave onDrawTitleBox unset.
+        nodeType.prototype.onDrawTitleBox = function (ctx, titleHeight, _size, scale) {
+            // Zoomed out far enough that LiteGraph itself drops to flat shapes,
+            // the mark is a few muddy pixels. Leave the title bar clean instead.
+            if (!markReady || (scale !== undefined && scale < 0.5)) {
+                return
+            }
+            const height = titleHeight * 0.62;
+            const width = height * MARK_ASPECT;
+            ctx.drawImage(
+                mark,
+                titleHeight * 0.5 - width * 0.5,
+                titleHeight * -0.5 - height * 0.5,
+                width,
+                height,
+            );
+        };
     },
     async setup() {
         console.info("vLLM-Omni Setup complete!")

apps/ComfyUI-vLLM-Omni/web/vllm-omni-mark.svg

@@ -0,0 +1,5 @@
+<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 157 165" width="157" height="165">
+  <title>vLLM-Omni</title>
+  <polygon points="0,36 62,36 62,160" fill="#F1B844"/>
+  <polygon points="96,36 157,0 121,165 68,165" fill="#54A0F8"/>
+</svg>

benchmarks/README.md

@@ -4,6 +4,11 @@ This directory contains benchmark suites for evaluating different model families
 
 ## Benchmark families
 
+### [Qwen2.5-Omni](qwen2_5_omni/README.md) — Embedding Wrapper Latency
+
+CPU-observed A/B timings for redundant text embedding removal, with real checkpoint
+embeddings, warmup, paired blocks, and embedding call-count validation.
+
 ### [TTS](tts/README.md) — Text-to-Speech
 
 Model-agnostic serving benchmarks for TTS models, including Qwen3-TTS and VoxCPM2.
@@ -36,6 +41,17 @@ LingBot-Video MoE transformer.
 - **Dense pipeline**: decoded-video MAE, MSE, PSNR, latency, and optional steady-state timings
 - **MoE transformer**: bitwise router, sparse-block, shared-expert, and full-transformer parity
 
+### [LingBot-World](lingbot_world/README.md) — Realtime Streaming Cadence
+
+Client-side benchmark for a `WS /v1/realtime/video` world-model session, where one
+request is a whole rollout and one AR block streams out as one video chunk. It
+measures a cadence rather than a throughput: there is exactly one session, and what
+matters is whether the next chunk arrives before the viewer finishes the last one.
+
+- **Layout**: `lingbot_world/benchmark_lingbot_world_realtime.py` (WS client), `lingbot_world/workload.py` (rollout and metric math), `lingbot_world/configs/` (single-GPU eager and Ulysses-4 compiled deploy configs)
+- **Workload**: built-in image-conditioned rollout with a generated camera script, or a JSON spec with per-chunk actions and mid-rollout prompt updates
+- **Key metrics**: TTFC (time to first chunk), steady-state chunk inter-arrival percentiles, VIDEO_RTF (wall seconds per video second; lower is better), chunk-deadline attainment, and simulated playback underruns
+
 ### [Distributed](distributed/omni_connectors/README.md) — RDMA Connector Testing
 
 RDMA environment setup and transfer tests for `MooncakeTransferEngineConnector`, including pytest-based single-node checks and manual cross-node benchmarks.
@@ -50,6 +66,13 @@ Accuracy benchmarks for image generation/editing models, adapting external suite
 - **Layout**: `accuracy/text_to_image/` (GEBench), `accuracy/image_to_image/` (GEdit-Bench)
 - **Method**: generation and judge scoring both run through local `vllm-omni serve` endpoints
 
+### [MammothModa2](mammoth_moda2/README.md) — Startup and Weight Loading
+
+Startup / model-loading benchmark for the two-stage MammothModa2 (AR → DiT) deployment, plus a raw safetensors loading micro-benchmark.
+
+- **Layout**: `mammoth_moda2/bench_startup.py` (single-process startup and first/steady request timing), `mammoth_moda2/parse_startup_log.py` (per-stage breakdown from engine logs), `mammoth_moda2/bench_storage_scenarios.sh` (cold/warm page cache on local disk vs. network storage), `mammoth_moda2/raw_load_bench.py` (safetensors → GPU without vLLM)
+- **Key metrics**: time to engine ready, per-stage spawn/init/weight-load/profile time, first vs. steady-state request latency
+
 ### Common serving metrics framework
 
 `vllm_omni/benchmarks/` extends `vllm bench serve --omni` with Omni-specific datasets, backends, and multimodal metrics. Key metrics include:

benchmarks/accuracy/common.py

@@ -1,3 +1,6 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+
 from __future__ import annotations
 
 import base64
@@ -200,7 +203,16 @@ class VllmOmniImageClient:
             )
             last_response = response
             if response.status_code != 404:
-                response.raise_for_status()
+                try:
+                    response.raise_for_status()
+                except requests.HTTPError as error:
+                    # Keep the server's validation detail in the exception;
+                    # otherwise multipart image tests only report a bare
+                    # ``400 Bad Request`` and hide the actionable cause.
+                    raise requests.HTTPError(
+                        f"{error}: {response.text}",
+                        response=response,
+                    ) from error
                 return decode_base64_image(response.json()["data"][0]["b64_json"])
 
         assert last_response is not None

benchmarks/accuracy/image_to_image/README.md

@@ -16,9 +16,9 @@ What changed:
   `/v1/images/edits` endpoint to produce benchmark outputs in the expected
   directory structure.
 - The evaluator keeps the same VIEScore-style decomposition:
-  - `semantics_score`
-  - `quality_score`
-  - `overall_score = sqrt(semantics_score * quality_score)`
+    - `semantics_score`
+    - `quality_score`
+    - `overall_score = sqrt(semantics_score * quality_score)`
 - Judge calls are routed to a local OpenAI-compatible model served by
   `vllm-omni`, not a remote provider.