Add support for OpenAI API o1 model (#3363)

Co-authored-by: Shan Yu <shanyu1@g.ucla.edu>

python/sglang/lang/backend/openai.py

@@ -161,6 +161,10 @@ class OpenAI(BaseBackend):
                 prompt = s.text_
 
             kwargs = sampling_params.to_openai_kwargs()
+            if self.model_name.startswith("o1") or self.model_name.startswith("o3"):
+                kwargs.pop("max_tokens", None)
+            else:
+                kwargs.pop("max_completion_tokens", None)
             comp = openai_completion(
                 client=self.client,
                 token_usage=self.token_usage,
@@ -175,6 +179,7 @@ class OpenAI(BaseBackend):
             ), "constrained type not supported on chat model"
             kwargs = sampling_params.to_openai_kwargs()
             kwargs.pop("stop")
+
             comp = openai_completion(
                 client=self.client,
                 token_usage=self.token_usage,

python/sglang/lang/ir.py

@@ -63,6 +63,7 @@ class SglSamplingParams:
             warnings.warn("Regular expression is not supported in the OpenAI backend.")
         return {
             "max_tokens": self.max_new_tokens,
+            "max_completion_tokens": self.max_new_tokens,
             "stop": self.stop or None,
             "temperature": self.temperature,
             "top_p": self.top_p,