diff --git a/.github/configs/nvidia-master.yaml b/.github/configs/nvidia-master.yaml
index 0eba30478c..989a56093e 100644
--- a/.github/configs/nvidia-master.yaml
+++ b/.github/configs/nvidia-master.yaml
@@ -2953,7 +2953,7 @@ dsv4-fp8-h200-vllm-agentic:
 # --speculative-config '{"method":"mtp","num_speculative_tokens":2}'.
 
 dsv4-fp8-h200-sglang:
-  image: lmsysorg/sglang:deepseek-v4-hopper@sha256:7f19c6dc092e47a10fac2e41f47eab78970280d06648b8e50d312a82f0ae722f
+  image: lmsysorg/sglang:v0.5.12-cu130
   model: deepseek-ai/DeepSeek-V4-Pro
   model-prefix: dsv4
   runner: h200-dgxc
@@ -2977,7 +2977,7 @@ dsv4-fp8-h200-sglang:
 # runner pool, search space) and adds EAGLE speculative decoding via
 # --speculative-algorithm EAGLE with the (3,1,4) chain matching dsv4-fp4-b300-sglang-mtp.
 dsv4-fp8-h200-sglang-mtp:
-  image: lmsysorg/sglang:deepseek-v4-hopper@sha256:7f19c6dc092e47a10fac2e41f47eab78970280d06648b8e50d312a82f0ae722f
+  image: lmsysorg/sglang:v0.5.12-cu130
   model: deepseek-ai/DeepSeek-V4-Pro
   model-prefix: dsv4
   runner: h200-dgxc
diff --git a/benchmarks/single_node/dsv4_fp8_h200_sglang.sh b/benchmarks/single_node/dsv4_fp8_h200_sglang.sh
index bf5c6f7b21..a669528790 100644
--- a/benchmarks/single_node/dsv4_fp8_h200_sglang.sh
+++ b/benchmarks/single_node/dsv4_fp8_h200_sglang.sh
@@ -43,7 +43,7 @@ PYTHONNOUSERSITE=1 sglang serve \
     --chunked-prefill-size 4096 \
     --disable-flashinfer-autotune \
     --disable-radix-cache \
-    --mem-fraction-static 0.88 \
+    --mem-fraction-static 0.92 \
     --max-running-requests "$(( CONC * 3 / 2 > 8 ? CONC * 3 / 2 : 8 ))" \
     $EVAL_CONTEXT_ARGS >> $SERVER_LOG 2>&1 &
 
diff --git a/benchmarks/single_node/dsv4_fp8_h200_sglang_mtp.sh b/benchmarks/single_node/dsv4_fp8_h200_sglang_mtp.sh
index bcba415435..d50c184d7d 100644
--- a/benchmarks/single_node/dsv4_fp8_h200_sglang_mtp.sh
+++ b/benchmarks/single_node/dsv4_fp8_h200_sglang_mtp.sh
@@ -43,7 +43,7 @@ PYTHONNOUSERSITE=1 sglang serve \
     --chunked-prefill-size 4096 \
     --disable-flashinfer-autotune \
     --disable-radix-cache \
-    --mem-fraction-static 0.88 \
+    --mem-fraction-static 0.92 \
     --max-running-requests "$(( CONC * 3 / 2 > 8 ? CONC * 3 / 2 : 8 ))" \
     --speculative-algorithm EAGLE \
     --speculative-num-steps 3 \
diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index ea86a018a6..a42b8ef9ee 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -3083,3 +3083,10 @@
   description:
     - "Update SGLang image from v0.5.10.post1-cu130 / v0.5.11-cu130 (30d old) to v0.5.12-cu130"
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1451
+
+- config-keys:
+    - dsv4-fp8-h200-sglang
+    - dsv4-fp8-h200-sglang-mtp
+  description:
+    - "Update SGLang image from SHA-pinned deepseek-v4-hopper custom build (15/14d old) to v0.5.12-cu130"
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1460