[Bugfix] fix precision issues of qwen3-omni when enable async_chunk without system prompt

vllm_omni/entrypoints/chat_utils.py

@@ -22,6 +22,9 @@
     _postprocess_messages,
     _ToolParser,
 )
+from vllm.logger import init_logger
+
+logger = init_logger(__name__)
 
 
 class OmniAsyncMultiModalItemTracker(AsyncMultiModalItemTracker):
@@ -135,6 +138,48 @@ def _cleanup_file_sync(file_path: str) -> None:
                 await asyncio.to_thread(_cleanup_file_sync, temp_video_file_path)
 
 
+def _ensure_system_prompt(
+    messages: list[ChatCompletionMessageParam],
+    model_config: ModelConfig,
+) -> list[ChatCompletionMessageParam]:
+    """
+    Ensure a system prompt exists for Qwen-Omni models to preserve precision of the model.
+    Args:
+        messages: List of chat messages
+        model_config: Model configuration
+
+    Returns:
+        Messages list with system prompt
+    """
+    model_name = getattr(model_config, "model", "").lower()
+    hf_config = getattr(model_config, "hf_config", None)
+    architectures = getattr(hf_config, "architectures", []) if hf_config else []
+
+    is_qwen_omni = ("qwen" in model_name and "omni" in model_name) or any(
+        "qwen" in arch.lower() and "omni" in arch.lower() for arch in architectures
+    )
+
+    if not is_qwen_omni:
+        return messages
+
+    if messages and messages[0].get("role") == "system":
+        return messages
+
+    default_qwen_omni_system_prompt = (
+        "You are Qwen, a virtual human developed by the Qwen Team, Alibaba "
+        "Group, capable of perceiving auditory and visual inputs, as well as "
+        "generating text and speech."
+    )
+
+    system_message: ChatCompletionMessageParam = {
+        "role": "system",
+        "content": default_qwen_omni_system_prompt,
+    }
+
+    logger.info(f"injecting system prompt {default_qwen_omni_system_prompt} for Qwen-Omni model")
+    return [system_message] + list(messages)
+
+
 def parse_chat_messages_futures(
     messages: list[ChatCompletionMessageParam],
     model_config: ModelConfig,
@@ -153,6 +198,9 @@ def parse_chat_messages_futures(
         Tuple of (conversation, mm_future) where mm_future resolves to
         (mm_data, mm_uuids) when awaited.
     """
+    # auto-inject system prompt for Qwen-Omni models if missing
+    messages = _ensure_system_prompt(messages, model_config)
+
     conversation: list[ConversationMessage] = []
     mm_tracker = OmniAsyncMultiModalItemTracker(model_config)
 

vllm_omni/model_executor/models/qwen3_omni/qwen3_omni.py

@@ -589,6 +589,7 @@ def talker_preprocess(self, input_ids: torch.Tensor, input_embeds: torch.Tensor,
             input_embeds = self.talker.embed_input_ids(input_ids)
 
         span_len = input_ids.shape[0]
+        update_dict = {}
         if span_len > 1:
             # prefill
             input_ids, input_embeds, update_dict = self.talker_preprocess_prefill(input_ids, input_embeds, **info_dict)
@@ -600,11 +601,12 @@ def talker_preprocess(self, input_ids: torch.Tensor, input_embeds: torch.Tensor,
             update_dict["code_predictor_codes"] = code_predictor_codes
         else:
             # decode
-            if info_dict.get("num_processed_tokens", 0) < len(info_dict.get("thinker_input_ids", [])):
+            if not info_dict.get("decode_flag", False):
                 info_dict["num_processed_tokens"] = len(info_dict.get("thinker_input_ids", [])) + 1
+                update_dict["decode_flag"] = True
 
             last_talker_hidden, text_step, update_dict = self.talker_preprocess_decode(
-                input_ids, input_embeds, **info_dict
+                input_ids, input_embeds, update_dict, **info_dict
             )
             update_dict["mtp_inputs"] = last_talker_hidden, text_step
 
@@ -874,8 +876,9 @@ def _thinker_decode_to_talker_decode(
         thinker_embed = thinker_embed[start_index : start_index + 1].to(device)
         return self.talker.text_projection(thinker_embed).to(device)
 
-    def talker_preprocess_decode(self, input_ids: torch.Tensor, input_embeds: torch.Tensor, **info_dict: dict):
-        update_dict: dict[str, dict] = {}
+    def talker_preprocess_decode(
+        self, input_ids: torch.Tensor, input_embeds: torch.Tensor, update_dict: dict, **info_dict: dict
+    ):
         last_talker_hidden = None
         text_step = None
         try: