feat: Add 'none' reasoning effort to ChatCompletionRequest

python/sglang/srt/entrypoints/openai/protocol.py

@@ -588,12 +588,13 @@ class ChatCompletionRequest(BaseModel):
     return_hidden_states: bool = False
     return_routed_experts: bool = False
     return_cached_tokens_details: bool = False
-    reasoning_effort: Optional[Literal["low", "medium", "high"]] = Field(
+    reasoning_effort: Optional[Literal["none", "low", "medium", "high"]] = Field(
         default="medium",
         description="Constrains effort on reasoning for reasoning models. "
-        "'low' is the least effort, 'high' is the most effort. Reducing reasoning effort can "
-        "result in faster responses and fewer tokens used on reasoning in a response. "
-        "Currently only supported for OpenAI models in the harmony path, i.e GPT-OSS models.",
+        "'none' disables reasoning entirely, 'low' is the least effort, 'high' is the most effort. "
+        "Reducing reasoning effort can result in faster responses and fewer tokens used on reasoning "
+        "in a response. 'none' defaults thinking and enable_thinking to false in "
+        "chat_template_kwargs (unless explicitly overridden). Not supported in the harmony path.",
     )
 
     # Extra parameters for SRT backend only and will be ignored by OpenAI models.
@@ -672,12 +673,10 @@ def set_tool_choice_default(cls, values):
     @classmethod
     def normalize_reasoning_inputs(cls, values: Dict):
         r = values.get("reasoning")
-        if r is None:
-            return values
 
-        if isinstance(r, dict):
+        if r is not None and isinstance(r, dict):
             effort = r.get("effort") or r.get("reasoning_effort")
-            if effort in {"low", "medium", "high"}:
+            if effort in {"none", "low", "medium", "high"}:
                 values["reasoning_effort"] = effort
 
             enabled = (
@@ -694,6 +693,17 @@ def normalize_reasoning_inputs(cls, values: Dict):
                 ctk.setdefault("thinking", True)
                 values["chat_template_kwargs"] = ctk
 
+        if values.get("reasoning_effort") == "none":
+            ctk = values.get("chat_template_kwargs")
+            if not isinstance(ctk, dict):
+                ctk = {}
+            # different models check different keys:
+            # - "thinking" for deepseek-v3, kimi_k2
+            # - "enable_thinking" for qwen3, glm45, nemotron_3, interns1
+            ctk.setdefault("thinking", False)
+            ctk.setdefault("enable_thinking", False)
+            values["chat_template_kwargs"] = ctk
+
         return values
 
     @model_validator(mode="before")

python/sglang/srt/entrypoints/openai/serving_chat.py

@@ -248,6 +248,11 @@ def _convert_to_internal_request(
             if request.chat_template_kwargs
             else None
         )
+        if self.is_gpt_oss and reasoning_effort == "none":
+            raise ValueError(
+                f"Harmony does not support reasoning effort {reasoning_effort}"
+            )
+
         if reasoning_effort is not None:
             request.reasoning_effort = reasoning_effort
 

test/registered/openai_server/basic/test_protocol.py

@@ -192,6 +192,30 @@ def test_chat_completion_reasoning_effort(self):
         self.assertEqual(request.reasoning_effort, "high")
         self.assertEqual(request.chat_template_kwargs, {"thinking": True})
 
+    def test_chat_completion_reasoning_effort_none(self):
+        """Test reasoning_effort='none' disables thinking"""
+        messages = [{"role": "user", "content": "Hello"}]
+        request = ChatCompletionRequest(
+            model="test-model",
+            messages=messages,
+            reasoning_effort="none",
+        )
+        self.assertEqual(request.reasoning_effort, "none")
+        self.assertFalse(request.chat_template_kwargs.get("thinking"))
+        self.assertFalse(request.chat_template_kwargs.get("enable_thinking"))
+
+    def test_chat_completion_reasoning_effort_none_from_reasoning_dict(self):
+        """Test reasoning_effort='none' via nested reasoning dict"""
+        messages = [{"role": "user", "content": "Hello"}]
+        request = ChatCompletionRequest(
+            model="test-model",
+            messages=messages,
+            reasoning={"effort": "none"},
+        )
+        self.assertEqual(request.reasoning_effort, "none")
+        self.assertFalse(request.chat_template_kwargs.get("thinking"))
+        self.assertFalse(request.chat_template_kwargs.get("enable_thinking"))
+
     def test_chat_completion_json_format(self):
         """Test chat completion json format"""
         transcript = "Good morning! It's 7:00 AM, and I'm just waking up. Today is going to be a busy day, "