[oai serving chat] Add argument --sampling-defaults and fix ChatCompletionRequest defaults (#11304)

python/sglang/srt/server_args.py

@@ -252,6 +252,7 @@ class ServerArgs:
     reasoning_parser: Optional[str] = None
     tool_call_parser: Optional[str] = None
     tool_server: Optional[str] = None
+    sampling_defaults: str = "model"
 
     # Data parallelism
     dp_size: int = 1
@@ -1872,6 +1873,16 @@ class ServerArgs:
             default=ServerArgs.tool_call_parser,
             help=f"Specify the parser for handling tool-call interactions. Options include: {tool_call_parser_choices}.",
         )
+        parser.add_argument(
+            "--sampling-defaults",
+            type=str,
+            choices=["openai", "model"],
+            default=ServerArgs.sampling_defaults,
+            help="Where to get default sampling parameters. "
+            "'openai' uses SGLang/OpenAI defaults (temperature=1.0, top_p=1.0, etc.). "
+            "'model' uses the model's generation_config.json to get the recommended "
+            "sampling parameters if available. Default is 'model'.",
+        )
         parser.add_argument(
             "--tool-server",
             type=str,