From a5bbcda968b69c89e8943d6dfdb6de8a3dd6a8d9 Mon Sep 17 00:00:00 2001 From: Yingchun Lai Date: Thu, 22 Jan 2026 12:22:20 +0800 Subject: [PATCH] fix: prefer to use max_completion_tokens rather than max_tokens (#17516) --- python/sglang/srt/entrypoints/openai/protocol.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/python/sglang/srt/entrypoints/openai/protocol.py b/python/sglang/srt/entrypoints/openai/protocol.py index 192852c9a..060e9659a 100644 --- a/python/sglang/srt/entrypoints/openai/protocol.py +++ b/python/sglang/srt/entrypoints/openai/protocol.py @@ -654,7 +654,7 @@ class ChatCompletionRequest(BaseModel): sampling_params = { "temperature": get_param("temperature"), - "max_new_tokens": self.max_tokens or self.max_completion_tokens, + "max_new_tokens": self.max_completion_tokens or self.max_tokens, "min_new_tokens": self.min_tokens, "stop": stop, "stop_token_ids": self.stop_token_ids,