[Docs] Unify recipe serve commands on vllm serve --omni (#6221)

Signed-off-by: Yueqian Lin <linyueqian@outlook.com>

recipes/BosonAI/Higgs-Audio-V3-TTS.md

@@ -45,7 +45,7 @@ MusicGen-style delay pattern).
 **Online serving:**
 
 ```bash
-vllm-omni serve bosonai/higgs-audio-v3-tts-4b \
+vllm serve bosonai/higgs-audio-v3-tts-4b \
     --host 0.0.0.0 --port 8095 \
     --trust-remote-code --omni
 ```

recipes/NVIDIA/Nemotron-Labs-Audex.md

@@ -51,7 +51,7 @@ NemotronH (~3B active parameters). Pick the pipeline by task:
 `model_type` auto-resolves `audex_tts.yaml`):
 
 ```bash
-vllm-omni serve nvidia/Nemotron-Labs-Audex-2B \
+vllm serve nvidia/Nemotron-Labs-Audex-2B \
     --host 0.0.0.0 --port 8097 \
     --trust-remote-code --omni
 # other modes: add --stage-configs-path vllm_omni/deploy/audex_{tta,thinker_only,s2s}.yaml
@@ -109,7 +109,7 @@ The 30B REQUIRES its explicit deploy yaml (the shared HF `model_type`
 auto-resolves to the 2B-tuned config otherwise):
 
 ```bash
-vllm-omni serve nvidia/Nemotron-Labs-Audex-30B-A3B \
+vllm serve nvidia/Nemotron-Labs-Audex-30B-A3B \
     --host 0.0.0.0 --port 8097 \
     --trust-remote-code --omni \
     --stage-configs-path vllm_omni/deploy/audex_tts_30b.yaml

recipes/StabilityAI/Stable-Audio-Open.md

@@ -78,7 +78,7 @@ python examples/offline_inference/text_to_audio/text_to_audio.py \
 Start the online serving endpoint:
 
 ```bash
-vllm-omni serve /path/to/stable-audio-open-1.0 \
+vllm serve /path/to/stable-audio-open-1.0 \
   --host 0.0.0.0 \
   --port 8091 \
   --gpu-memory-utilization 0.9 \

recipes/TEMPLATE.md

@@ -39,6 +39,9 @@ platform section, document one or more tested hardware configurations.
 
 #### Command
 
+Serve commands use `vllm serve <model> --omni`. The `--omni` flag is what
+selects the Omni pipeline, so it is required and cannot be dropped.
+
 ```bash
 # Add the exact command(s) here
 ```

recipes/inclusionAI/Ming-omni-tts.md

@@ -39,7 +39,7 @@ Other hardware is welcome as community validation lands.
 Launch the two-stage talker:
 
 ```bash
-vllm-omni serve inclusionAI/Ming-omni-tts-0.5B \
+vllm serve inclusionAI/Ming-omni-tts-0.5B \
     --deploy-config vllm_omni/deploy/ming_tts.yaml \
     --omni \
     --port 8091 \

recipes/zai-org/GLM-TTS.md

@@ -44,20 +44,20 @@ conditioned on reference audio and its transcript.
 Start the server from the repository root:
 
 ```bash
-vllm-omni serve zai-org/GLM-TTS --omni --trust-remote-code --port 8091
+vllm serve zai-org/GLM-TTS --omni --trust-remote-code --port 8091
 ```
 
 Async chunking is enabled by default in the bundled deployment config. For
 the sync (non-streaming) path:
 
 ```bash
-vllm-omni serve zai-org/GLM-TTS --omni --trust-remote-code --port 8091 --no-async-chunk
+vllm serve zai-org/GLM-TTS --omni --trust-remote-code --port 8091 --no-async-chunk
 ```
 
 Use a custom deploy config for advanced cases:
 
 ```bash
-vllm-omni serve zai-org/GLM-TTS --omni --trust-remote-code --port 8091 \
+vllm serve zai-org/GLM-TTS --omni --trust-remote-code --port 8091 \
   --deploy-config /path/to/your_glm_tts_overrides.yaml
 ```