fix(openai): avoid duplicate routed expert in response when return_meta_info = True (#35323)

2026-08-20 15:21:58 -07:00
parent 94907f05c4
commit a4ef828207
3 changed files with 110 additions and 5 deletions
@@ -1819,7 +1819,11 @@ class OpenAIServingChat(OpenAIServingBase):

        # Build sglext at response level (from first ret_item, as these are per-request)
        first_ret = ret[0]
-        routed_experts = process_routed_experts_from_ret(first_ret, request)
+        routed_experts = (
+            None
+            if request.return_meta_info
+            else process_routed_experts_from_ret(first_ret, request)
+        )
        cached_tokens_details = process_cached_tokens_details_from_ret(
            first_ret, request
        )