Merge branch 'main' into qwen-image-2.1

Signed-off-by: Isotr0py <Isotr0py@outlook.com>

docs/configuration/environment_variables.md

@@ -88,6 +88,7 @@ depends on the installed kernels and model path.
 | `VLLM_OMNI_EVENT_DRIVEN_ORCH` | `1`, `true`, `yes` or `on` enables; defaults to on for Qwen3-TTS and off for other pipelines | Pipeline default computed from `pipeline_config.model_type` at engine initialization; env override resolved at `Orchestrator` construction and separately when the serving-side final-output drain starts | An explicit env value wins; otherwise both consumers use the engine's pipeline default. Values are stripped and case-normalized; any unrecognized value selects the legacy poll loop. Set before server startup. | Experimental |
 | `VLLM_OMNI_INPUT_WAIT_TIMEOUT_S` | Float seconds; default `600`; `<=0` disables | Full-payload input coordinator, not async-chunk transfer; read when the scheduler module imports in each worker | Environment-only setting. A non-float logs a warning and uses `600`. | Stable operational control |
 | `VLLM_OMNI_ORCH_MONITOR_PATH` | Filesystem path; default `<current-working-directory>/vllm_omni_orch_monitor_<timestamp>.json` | Orchestrator monitor enabled by `--enable-orch-monitor`; read when the monitor is created | Environment-only path override. Parent directories are created; write errors are logged. | Diagnostic |
+| `VLLM_OMNI_SPEAKER_REGISTRATION_POLICY` | `overwrite` or `immutable`; default `overwrite` | Speech server; read when speaker storage initializes | Environment-only setting. `immutable` rejects re-registering an existing uploaded voice name until it is deleted; any other value raises `ValueError` at startup. | Experimental |
 | `VLLM_OMNI_VIDEO_SYNC_TIMEOUT` | Float seconds; default `600` | Synchronous Videos API; read when the API server module imports | Environment-only setting. A non-float raises `ValueError` during import. | Experimental |
 | `VLLM_VIDEO_ASYNC_CHUNK` | `on` or `off`; default `on` | Streaming video output; read on attribute access | Environment-only setting. Values are trimmed and case-normalized; an invalid value warns once and uses `on`. | Experimental |
 | `VLLM_VIDEO_AUDIO_DELTA_MODE` | `fast` or `slow`; default `fast` | Streaming video audio deltas; read on attribute access | Environment-only setting. Values are trimmed and case-normalized; an invalid value warns once and uses `fast`. | Experimental |

docs/design/feature/cache_dit.md

@@ -180,17 +180,52 @@ cache_dit.enable_cache(
 
 > **Note:** For single transformer with multiple block lists, `refresh_context` works the same as standard models.
 
+### Example 3: Sequential-CFG Diffusion Pipeline (MammothModa2)
+
+MammothModa2's DiT stage runs on the diffusion runner, which enables the
+configured backend at startup and transfers ownership to the pipeline through
+the request-scoped protocol (`adopt_cache_dit_backend` / `is_cache_dit_enabled`);
+`forward()` then reconciles per-request state through
+`RequestScopedCacheDiTRuntime` (MiniMax H3 precedent).
+
+**Key difference:** The enabler caches only the repeated main-layer stack and
+marks the pipeline as requiring paired CFG forwards, because cache-dit's
+separate-CFG accounting distinguishes cond/uncond passes purely by forward
+parity:
+
+```python
+pipeline._cache_dit_requires_paired_cfg = True
+block_adapter = BlockAdapter(
+    transformer=pipeline.gen_transformer,
+    blocks=[pipeline.gen_transformer.layers],  # Refiners stay outside
+    forward_pattern=[ForwardPattern.Pattern_3],
+    has_separate_cfg=True,
+    check_forward_pattern=True,
+)
+```
+
+Per request, the pipeline calls `runtime.prepare(spec)` for CFG requests
+(refreshing the step context) and `runtime.prepare(None)` for `guidance=1.0`
+requests (disabling hooks — their single forward per step cannot be expressed
+by the parity accounting).
+
 ### Registering Custom Implementations
 
-After writing your custom enabler, register it in `CUSTOM_DIT_ENABLERS` in
+After writing your custom enabler, register it in
+`CUSTOM_DIT_ENABLERS` via `register_custom_dit_enablers()` in
 `vllm_omni/diffusion/cache/cachedit/model_specific.py`:
 
 ```python
-CUSTOM_DIT_ENABLERS = {
-    "Wan22Pipeline": enable_cache_for_wan22,
-    "LongCatImagePipeline": enable_cache_for_longcat_image,
-    "YourCustomPipeline": enable_cache_for_your_model,  # Add here
-}
+CUSTOM_DIT_ENABLERS.update(
+    {
+        "Wan22Pipeline": enable_cache_for_wan22,
+        "Cosmos3OmniDiffusersPipeline": enable_cache_for_cosmos3,
+        "Krea2Pipeline": enable_cache_for_krea2,
+        "Magi2Pipeline": enable_cache_for_magi2,
+        "MammothModa2DiTPipeline": enable_cache_for_mammothmoda2,
+        "YourCustomPipeline": enable_cache_for_your_model,  # Add here
+    }
+)
 ```
 
 ---
@@ -274,6 +309,7 @@ Complete examples in the codebase:
 | **Wan2.2** | `cachedit.model_specific::enable_cache_for_wan22` | Single or dual-transformer | Auto-detects mode based on transformer_2 presence |
 | **LongCat** | `cachedit.config::CacheDiTAdapterConfig` | Declarative block adapter | Two block lists in one transformer |
 | **BAGEL** | `cachedit.model_specific::BagelCachedAdapter` | Custom cached adapter | Complex architecture |
+| **MammothModa2** | `cachedit.model_specific::enable_cache_for_mammothmoda2` | Sub-stack + paired CFG | Generation-runner pipeline; lifecycle owned by `pipeline_mammothmoda2_dit.py` |
 
 ---
 

docs/serving/speech_api.md

@@ -289,6 +289,19 @@ Upload a new voice sample for voice cloning in Base task TTS requests.
 
 Fields `ref_text` and `speaker_description` are omitted when not provided at upload time.
 
+**Naming rules:**
+
+- Names that collide with one of the model's built-in or precomputed voices are rejected (400). Voice
+  files already on disk under such a name are ignored at startup with a warning, so an upload can never
+  shadow a built-in voice.
+- Re-uploading an existing uploaded name overwrites it in place (the previous audio file is
+  deleted). Set `VLLM_OMNI_SPEAKER_REGISTRATION_POLICY=immutable` on the server to reject duplicates instead,
+  requiring an explicit `DELETE /v1/audio/voices/{name}` before re-registering — useful when the
+  endpoint is reachable by multiple writers and silent overwrites are a risk.
+- The voice registry has no per-user ownership: any client that can reach the endpoint can
+  overwrite (default policy) or delete any uploaded voice. For multi-tenant deployments, add
+  authentication at a proxy and namespace voice names per user (e.g. `{user}.{name}`).
+
 **Usage Example:**
 
 ```bash
@@ -577,6 +590,7 @@ by `GET /v1/audio/voices`. Valid precomputed voices can be used in
 | ---------- | --------- | ------------- |
 | `SPEAKER_SAMPLES_DIR` | `~/.cache/vllm-omni/speakers` | Directory for persisted uploaded speakers (`.safetensors` files). |
 | `SPEAKER_MAX_UPLOADED` | `1000` | Maximum number of uploaded speakers kept on disk. Upload requests past the cap return 400. |
+| `VLLM_OMNI_SPEAKER_REGISTRATION_POLICY` | `overwrite` | `immutable` rejects re-uploading an existing uploaded name (400) until it is deleted; any other value fails startup. |
 
 The in-memory LRU has a fixed 512 MiB byte budget.
 

docs/user_guide/diffusion/cache_acceleration/cache_dit.md

@@ -72,6 +72,39 @@ omni = Omni(
 
 ## Example Script
 
+### Multi-Stage Models (MammothModa2)
+
+MammothModa2 is a multi-stage AR→DiT pipeline; the diffusion runner owns the
+Cache-DiT lifecycle of its DiT stage. Configure it on the **DiT stage entry**
+of the deploy YAML so the acceleration only applies to the denoising stage:
+
+```yaml
+# deploy YAML (see vllm_omni/deploy/mammoth_moda2.yaml)
+stages:
+  - stage_id: 1
+    # ... other DiT stage settings ...
+    cache_backend: cache_dit
+    cache_config:
+      Fn_compute_blocks: 1
+      Bn_compute_blocks: 0
+      max_warmup_steps: 4
+      residual_diff_threshold: 0.24
+      max_continuous_cached_steps: 3
+    enable_cache_dit_summary: true   # log skip-ratio stats per request
+```
+
+Behavior notes:
+
+- Only the repeated main-layer stack is cached; the Q-Former refiners always run.
+- Sequential-CFG parity requires the unconditional pass on every denoise step,
+  so the `cfg_range` skip optimization is not used: outside the interval CFG is
+  neutralized with `scale=1.0` instead of skipping the uncond forward.
+- Requests with `text_guidance_scale = 1.0` run with cache hooks disabled.
+- `residual_diff_threshold` defaults to `0.24` (measured ~1.7-2.2x at 50 steps,
+  PSNR 27-34 dB vs uncached). For 20-step generation — especially with a
+  partial `cfg_range` — use `0.12` for noticeably closer parity at a modest
+  speed cost.
+
 ### Offline Inference
 
 Use the example script under `examples/offline_inference/text_to_image`:

docs/user_guide/diffusion_features.md

@@ -130,7 +130,7 @@ The following tables show which models support each feature:
 | **Krea 2**               |     ❌     |     ✅      |           ❌           |       ❌        |         ❌         |          ❌          |   ✅    |             ✅             |      ✅ (decode)      |       ❌        |        ❌         |
 | **LongCat-Image**        |     ✅     |     ✅      |           ✅           |       ✅        |         ✅         |          ❌          |   ❌    |             ✅             |          ❌           |       ❌        |        ❌         |
 | **LongCat-Image-Edit**   |     ✅     |     ✅      |           ✅           |       ✅        |         ✅         |          ❌          |   ❌    |             ✅             |          ❌           |       ❌        |        ❌         |
-| **MammothModa2(T2I)**    |     ❌     |     ❌      |           ❌           |       ❌        |         ❌         |          ❌          |   ❌    |             ❌             |          ❌           |       ❌        |        ❌         |
+| **MammothModa2(T2I)**    |     ❌     |     ✅      |           ❌           |       ❌        |         ❌         |          ❌          |   ❌    |             ❌             |          ❌           |       ❌        |        ❌         |
 | **Nextstep_1(T2I)**      |     ❓     |     ❓      |           ❌           |       ✅        |         ✅         |          ❌          |   ❌    |             ✅             |          ❌           |       ❌        |        ❌         |
 | **OmniGen2**             |     ❌     |     ✅      |           ✅           |       ❌        |         ✅         |          ❌          |   ❌    |             ❌             |          ❌           |       ❌        |        ❌         |
 | **Ovis-Image**           |     ❌     |     ✅      |           ❌           |       ✅        |         ❌         |          ❌          |   ❌    |             ✅             |          ❌           |       ❌        |        ❌         |
@@ -155,7 +155,8 @@ The following tables show which models support each feature:
 > 4. Krea 2 currently supports single-GPU inference plus LoRA, Cache-DiT, HSDP, CPU/layerwise offload, and VAE-patch-parallel (decode). TP/SP/CFG-Parallel are not yet wired. The few-step distilled (Turbo) checkpoint uses `is_distilled=true` (fixed timestep shift `mu=1.15`); generate at 2048x2048 by default with `num_inference_steps≈8` and `guidance_scale=0`. The Raw checkpoint uses 1024x1024, `num_inference_steps=28`, and `guidance_scale=4.5`.
 > 5. HunyuanImage3 supports step execution. Multi-request step execution requires `TORCH_SDPA`; see [Diffusion Execution Modes](diffusion/execution_modes.md#step-execution).
 > 6. BAGEL step execution supports image generation with `bagel.yaml`, `bagel_think.yaml`, and `bagel_single_stage.yaml`; two-stage Thinker execution and explicit single-stage text output remain on their existing complete-request paths. Image requests require `num_inference_steps >= 2`. BAGEL step execution cannot currently be combined with sequence parallelism or a diffusion cache backend; see [Diffusion Execution Modes](diffusion/execution_modes.md#step-execution).
-> 7. Qwen-Image-2.1 (`QwenImage21Pipeline`) handles both text-to-image and image-conditioned generation (up to 4 condition images) and supports LoRA through `QwenImageLoraLoaderMixin`. Sequence parallelism is Ulysses-only (ring attention is not supported), and cache acceleration (TeaCache / Cache-DiT) is unavailable; its built-in KV cache for the text/condition prefix is a separate feature from those cache backends.
+> 7. MammothModa2 runs its DiT stage on the diffusion runner (`StageExecutionType.DIFFUSION`); Cache-DiT is enabled through the standard diffusion-stage knobs on the stage entry of the deploy YAML (`cache_backend: cache_dit`, plus optional `cache_config` / `enable_cache_dit_summary`). The runner installs the backend at startup and the pipeline adopts it per request. Only the repeated main-layer stack is cached; requests with `text_guidance_scale = 1.0` bypass the cache hooks.
+> 8. Qwen-Image-2.1 (`QwenImage21Pipeline`) handles both text-to-image and image-conditioned generation (up to 4 condition images) and supports LoRA through `QwenImageLoraLoaderMixin`. Sequence parallelism is Ulysses-only (ring attention is not supported), and cache acceleration (TeaCache / Cache-DiT) is unavailable; its built-in KV cache for the text/condition prefix is a separate feature from those cache backends.
 
 ### VideoGen
 

docs/user_guide/examples/offline_inference/text_to_speech.md

@@ -2,7 +2,6 @@
 
 Source <https://github.com/vllm-project/vllm-omni/tree/main/examples/offline_inference/text_to_speech>.
 
-
 vLLM-Omni supports several autoregressive TTS models. They share a common
 CLI shape (`--text`, `--ref-audio`, `--ref-text`, `--output-dir`) and live
 together in this hub. Each model has its own subdirectory containing a
@@ -15,7 +14,7 @@ list of supported architectures across all modalities, see
 ## Supported Models
 
 | Model | HuggingFace repo | Stages | Voice cloning | Streaming | Special modes | Sample rate |
-|---|---|---|---|---|---|---|
+| --- | --- | --- | --- | --- | --- | --- |
 | VoxCPM2 | `openbmb/VoxCPM2` | single (native AR) | ✓ | — | continuation (`--ref-audio` + `--ref-text`) | 48 kHz |
 | CosyVoice3 | `FunAudioLLM/Fun-CosyVoice3-0.5B-2512` | 2 (talker + code2wav) | ✓ | ✓ (`async_chunk: true` default) | — | 22.05 kHz |
 | Fish Speech S2 Pro | `fishaudio/s2-pro` | dual-AR | ✓ | ✓ (`--streaming`) | — | 44.1 kHz |
@@ -45,13 +44,15 @@ python examples/offline_inference/text_to_speech/<model>/end2end.py \
 Single-stage native AR TTS at 48 kHz. Pipeline: `feat_encoder → MiniCPM4 → FSQ → residual_lm → LocDiT → AudioVAE`.
 
 ### Prerequisites
+
 ```bash
 pip install voxcpm
 # or, for a local source checkout:
-export VLLM_OMNI_VOXCPM_CODE_PATH=/path/to/voxcpm
+pip install -e /path/to/voxcpm
 ```
 
 ### Quick start
+
 ```bash
 python examples/offline_inference/text_to_speech/voxcpm2/end2end.py \
     --model openbmb/VoxCPM2 \
@@ -59,7 +60,9 @@ python examples/offline_inference/text_to_speech/voxcpm2/end2end.py \
 ```
 
 ### Voice cloning
+
 Pass a reference audio for isolated cloning, or both `--ref-audio` + `--ref-text` for prompt continuation:
+
 ```bash
 python examples/offline_inference/text_to_speech/voxcpm2/end2end.py \
     --text "Hello, this is a voice clone demo." \
@@ -68,6 +71,7 @@ python examples/offline_inference/text_to_speech/voxcpm2/end2end.py \
 ```
 
 ### Notes
+
 - Output: 48 kHz mono WAV.
 - Deploy config: `vllm_omni/deploy/voxcpm2.yaml` (default).
 
@@ -78,12 +82,14 @@ python examples/offline_inference/text_to_speech/voxcpm2/end2end.py \
 2-stage TTS pipeline (`talker` + `code2wav`) at 22.05 kHz.
 
 ### Prerequisites
+
 ```bash
 uv pip install -e .
 # Includes soundfile, onnxruntime, x-transformers, einops via requirements.
 ```
 
 Download the model snapshot:
+
 ```python
 from huggingface_hub import snapshot_download
 snapshot_download('FunAudioLLM/Fun-CosyVoice3-0.5B-2512',
@@ -91,15 +97,18 @@ snapshot_download('FunAudioLLM/Fun-CosyVoice3-0.5B-2512',
 ```
 
 If your downloaded checkpoint lacks `config.json`, add it:
+
 ```json
 {
     "model_type": "cosyvoice3",
     "architectures": ["CosyVoice3Model"]
 }
 ```
+
 This is required because `AutoConfig.register("cosyvoice3", CosyVoice3Config)` only registers the class mapping; the loader still reads `model_type` from `config.json` to select the class.
 
 ### Quick start
+
 ```bash
 python examples/offline_inference/text_to_speech/cosyvoice3/end2end.py \
     --model pretrained_models/Fun-CosyVoice3-0.5B \
@@ -107,7 +116,9 @@ python examples/offline_inference/text_to_speech/cosyvoice3/end2end.py \
 ```
 
 ### Voice cloning
+
 Pass a reference audio. Note that CosyVoice3's `--prompt-text` is a system-style prompt for the GPT stage, not a reference transcript:
+
 ```bash
 python examples/offline_inference/text_to_speech/cosyvoice3/end2end.py \
     --model pretrained_models/Fun-CosyVoice3-0.5B \
@@ -117,6 +128,7 @@ python examples/offline_inference/text_to_speech/cosyvoice3/end2end.py \
 ```
 
 ### Notes
+
 - Stage 0 (`talker`) emits speech tokens; stage 1 (`code2wav`) runs flow matching + HiFiGAN to synthesize waveform.
 - Deploy config auto-loads from `vllm_omni/deploy/cosyvoice3.yaml` based on HF `model_type`. Pass `--deploy-config <path>` to override.
 - `async_chunk: true` is the default; pass `--no-async-chunk` to switch to the legacy synchronous path.
@@ -129,12 +141,14 @@ python examples/offline_inference/text_to_speech/cosyvoice3/end2end.py \
 No extra packages are required; the DAC codec is vendored in vLLM-Omni.
 
 ### Quick start
+
 ```bash
 python examples/offline_inference/text_to_speech/fish_speech/end2end.py \
     --text "Hello, this is a test of the Fish Speech text to speech system."
 ```
 
 ### Voice cloning
+
 ```bash
 python examples/offline_inference/text_to_speech/fish_speech/end2end.py \
     --text "Hello, this is a cloned voice." \
@@ -143,14 +157,17 @@ python examples/offline_inference/text_to_speech/fish_speech/end2end.py \
 ```
 
 ### Streaming
+
 ```bash
 python examples/offline_inference/text_to_speech/fish_speech/end2end.py \
     --text "Hello, this is a streaming test." \
     --streaming
 ```
+
 Streaming requires `async_chunk: true` in the deploy config.
 
 ### Notes
+
 - Output: 44.1 kHz mono WAV.
 - DAC codec weights (`codec.pth`) are loaded lazily from the model directory.
 
@@ -161,6 +178,7 @@ Streaming requires `async_chunk: true` in the deploy config.
 2-stage TTS pipeline (AR + DiT flow-matching) at 24 kHz. Every request requires reference audio and its transcript for zero-shot voice cloning.
 
 ### Quick start
+
 ```bash
 python examples/offline_inference/text_to_speech/glm_tts/end2end.py \
     --model zai-org/GLM-TTS \
@@ -171,12 +189,14 @@ python examples/offline_inference/text_to_speech/glm_tts/end2end.py \
 ```
 
 ### Architecture
-```
+
+```text
 Text → [Stage 0: AR] → Speech Tokens → [Stage 1: DiT + HiFT] → Audio (24 kHz)
         (Llama-based)    (32k vocab)      (Flow Matching)
 ```
 
 ### Notes
+
 - `--ref-audio` and `--ref-text` are **required** together; GLM-TTS does not support text-only synthesis.
 - Reference audio should be 3-10 seconds.
 - First run may be slow due to lazy loading of WhisperVQ tokenizer and CampPlus ONNX speaker embedder.
@@ -190,12 +210,15 @@ Text → [Stage 0: AR] → Speech Tokens → [Stage 1: DiT + HiFT] → Audio (24
 Zero-shot multilingual TTS supporting 600+ languages, with three modes (auto / clone / design).
 
 ### Prerequisites
+
 ```bash
 huggingface-cli download k2-fsa/OmniVoice
 ```
+
 Voice cloning requires `transformers>=5.3.0`. Auto and design modes work with `transformers>=4.57.0`.
 
 ### Quick start (auto voice)
+
 ```bash
 python examples/offline_inference/text_to_speech/omnivoice/end2end.py \
     --model k2-fsa/OmniVoice \
@@ -203,6 +226,7 @@ python examples/offline_inference/text_to_speech/omnivoice/end2end.py \
 ```
 
 ### Voice cloning
+
 ```bash
 python examples/offline_inference/text_to_speech/omnivoice/end2end.py \
     --model k2-fsa/OmniVoice \
@@ -212,6 +236,7 @@ python examples/offline_inference/text_to_speech/omnivoice/end2end.py \
 ```
 
 ### Voice design
+
 ```bash
 python examples/offline_inference/text_to_speech/omnivoice/end2end.py \
     --model k2-fsa/OmniVoice \
@@ -220,6 +245,7 @@ python examples/offline_inference/text_to_speech/omnivoice/end2end.py \
 ```
 
 ### Language hint
+
 ```bash
 python examples/offline_inference/text_to_speech/omnivoice/end2end.py \
     --model k2-fsa/OmniVoice \
@@ -228,6 +254,7 @@ python examples/offline_inference/text_to_speech/omnivoice/end2end.py \
 ```
 
 ### Notes
+
 - Stage 0 (Generator): Qwen3-0.6B with 32-step iterative unmasking.
 - Stage 1 (Decoder): HiggsAudioV2 RVQ + DAC at 24 kHz.
 
@@ -238,13 +265,16 @@ python examples/offline_inference/text_to_speech/omnivoice/end2end.py \
 3-task-variant TTS with 24 kHz output. Has its own argparse surface (this script does not follow the common `--text` / `--ref-audio` shape).
 
 ### Prerequisites
+
 For ROCm builds, replace `onnxruntime` with `onnxruntime-rocm`:
+
 ```bash
 pip uninstall onnxruntime
 pip install onnxruntime-rocm
 ```
 
 ### Task variants
+
 - `CustomVoice`: predefined speaker (speaker ID) with optional style instruction.
 - `VoiceDesign`: text + descriptive instruction designs a new voice.
 - `Base`: voice cloning from reference audio + transcript.
@@ -264,16 +294,20 @@ python examples/offline_inference/text_to_speech/qwen3_tts/end2end.py --query-ty
 ```
 
 ### Streaming
+
 ```bash
 python examples/offline_inference/text_to_speech/qwen3_tts/end2end.py \
     --query-type CustomVoice \
     --streaming \
     --output-dir /tmp/out_stream
 ```
+
 Streaming requires `async_chunk: true` in the deploy config.
 
 ### Batched decoding
+
 The Code2Wav stage supports batched decoding through the SpeechTokenizer. Configure both stages with `max_num_seqs > 1` via `--stage-overrides` and pass multiple prompts via `--txt-prompts`:
+
 ```bash
 python examples/offline_inference/text_to_speech/qwen3_tts/end2end.py \
     --query-type CustomVoice \
@@ -281,9 +315,11 @@ python examples/offline_inference/text_to_speech/qwen3_tts/end2end.py \
     --batch-size 4 \
     --stage-overrides '{"0":{"max_num_seqs":4,"gpu_memory_utilization":0.2},"1":{"max_num_seqs":4,"gpu_memory_utilization":0.2}}'
 ```
+
 `--batch-size` must match a CUDA-graph capture size (1, 2, 4, 8, 16…).
 
 ### Notes
+
 - Run `--help` for the full argument surface.
 - See `qwen3_tts/end2end.py` for the prompt-length-estimation logic the Talker uses.
 
@@ -294,12 +330,15 @@ python examples/offline_inference/text_to_speech/qwen3_tts/end2end.py \
 Voxtral-4B-TTS (Mistral). Has its own argparse surface; uses voice presets and the `mistral_common` `SpeechRequest` protocol.
 
 ### Prerequisites
+
 Latest `mistral_common` with `SpeechRequest` support:
+
 ```bash
 pip install -e /path/to/mistral-common  # or upgrade from PyPI when available
 ```
 
 ### Quick start (voice preset)
+
 ```bash
 python examples/offline_inference/text_to_speech/voxtral_tts/end2end.py \
     --write-audio --voice cheerful_female \
@@ -308,6 +347,7 @@ python examples/offline_inference/text_to_speech/voxtral_tts/end2end.py \
 ```
 
 ### Voice cloning (capability gated upstream)
+
 ```bash
 python examples/offline_inference/text_to_speech/voxtral_tts/end2end.py \
     --write-audio \
@@ -317,15 +357,18 @@ python examples/offline_inference/text_to_speech/voxtral_tts/end2end.py \
 ```
 
 ### Streaming + concurrency
+
 ```bash
 python examples/offline_inference/text_to_speech/voxtral_tts/end2end.py \
     --num-prompts 32 --concurrency 8 --streaming --write-audio --voice neutral_female \
     --model mistralai/Voxtral-4B-TTS-2603 \
     --text "..."
 ```
+
 Available voice presets are listed on the HF model card (`mistralai/Voxtral-4B-TTS-2603`).
 
 ### Notes
+
 - `--num-prompts N` replicates the prompt for performance measurement.
 - `--concurrency M` requires `--streaming` and must evenly divide `--num-prompts`.
 - Run `--help` for the full argument surface.