From c6a45fab649db965c0eddb620ebf8d1782304d17 Mon Sep 17 00:00:00 2001
From: Xiaoyu Zhang <35585791+BBuf@users.noreply.github.com>
Date: Sat, 18 Apr 2026 22:32:41 +0800
Subject: [PATCH] Qwen3next flashinfer allreduce auto enable (#22664)

---
 python/sglang/srt/server_args.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/python/sglang/srt/server_args.py b/python/sglang/srt/server_args.py
index 654d47659..642fc2a99 100644
--- a/python/sglang/srt/server_args.py
+++ b/python/sglang/srt/server_args.py
@@ -2179,7 +2179,8 @@ class ServerArgs:
             )
 
         # TRTLLM AllReduce Fusion supports SM90/100, enable it by default
-        # for models with explicit support (DeepseekV3, GptOss, Glm4Moe, Qwen3Moe)
+        # for models with explicit support (DeepseekV3, GptOss, Glm4Moe,
+        # Qwen3/Qwen3Next/Qwen3.5 MoE families)
         # TODO: currently, it is only supported in the single node scenario. https://github.com/flashinfer-ai/flashinfer/issues/2006
         # TODO: there is currently a bug on H20 device specifically, https://github.com/flashinfer-ai/flashinfer/issues/2204
         device_name = get_device_name()
@@ -2197,6 +2198,7 @@ class ServerArgs:
                 "Glm4MoeForCausalLM",
                 "Glm4MoeLiteForCausalLM",
                 "Qwen3MoeForCausalLM",
+                "Qwen3NextForCausalLM",
                 "KimiK25ForConditionalGeneration",
                 "Qwen3_5MoeForConditionalGeneration",
                 "Qwen3_5ForConditionalGeneration",