diff --git a/.github/configs/amd-master.yaml b/.github/configs/amd-master.yaml
index 89318004b..084893e11 100644
--- a/.github/configs/amd-master.yaml
+++ b/.github/configs/amd-master.yaml
@@ -242,7 +242,7 @@ qwen3.5-fp8-mi300x-sglang:
     - { tp: 8, conc-start: 4, conc-end: 64 }
 
 glm5-fp8-mi355x-sglang:
-  image: rocm/sgl-dev:v0.5.8.post1-rocm720-mi35x-20260219
+  image: lmsysorg/sglang-rocm:v0.5.10rc0-rocm720-mi35x-20260413
   model: zai-org/GLM-5-FP8
   model-prefix: glm5
   runner: mi355x
diff --git a/benchmarks/single_node/glm5_fp8_mi355x.sh b/benchmarks/single_node/glm5_fp8_mi355x.sh
index 3d82fd856..668120e57 100755
--- a/benchmarks/single_node/glm5_fp8_mi355x.sh
+++ b/benchmarks/single_node/glm5_fp8_mi355x.sh
@@ -49,7 +49,9 @@ python3 -m sglang.launch_server \
     --mem-fraction-static 0.85 \
     --model-loader-extra-config '{"enable_multithread_load": true, "num_threads": 8}' \
     --nsa-prefill-backend tilelang \
-    --nsa-decode-backend tilelang $EVAL_CONTEXT_ARGS > $SERVER_LOG 2>&1 &
+    --nsa-decode-backend tilelang $EVAL_CONTEXT_ARGS  \
+    --kv-cache-dtype fp8_e4m3 \
+    --disable-radix-cache> $SERVER_LOG 2>&1 &
 
 SERVER_PID=$!
 
diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index 6721dbb1e..73f0d86ce 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -1353,6 +1353,13 @@
     - "TP2/TP4 seach space exploration for Qwen3.5 fp4 on SGL"
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1022
   
+- config-keys:
+    - glm5-fp8-mi355x-sglang
+  description:
+    - "Upgrade GLM5 FP8 MI355X SGLang image to v0.5.10rc0-rocm720-mi35x-20260413"
+    - "Set --kv-cache-dtype fp8_e4m3 and --disable-radix-cache"
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1023
+
 - config-keys:
     - qwen3.5-fp8-h200-sglang-mtp
   description: