From 9191b02eda0e8a1117f092e75f514d608d540f99 Mon Sep 17 00:00:00 2001
From: weireweire <weiliangl@nvidia.com>
Date: Wed, 1 Apr 2026 06:20:56 +0800
Subject: [PATCH] Fix cuda graph max bs capture upper bound (#21005)

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

diff --git a/python/sglang/srt/server_args.py b/python/sglang/srt/server_args.py
index 3d185689d..ce4cc7df4 100644
--- a/python/sglang/srt/server_args.py
+++ b/python/sglang/srt/server_args.py
@@ -1352,6 +1352,9 @@ class ServerArgs:
 
         capture_bs = [bs for bs in capture_bs if bs <= self.cuda_graph_max_bs]
 
+        if self.cuda_graph_max_bs not in capture_bs:
+            capture_bs.append(self.cuda_graph_max_bs)
+
         return capture_bs
 
     def _generate_cpu_graph_batch_sizes(self):