diff --git a/.github/configs/nvidia-master.yaml b/.github/configs/nvidia-master.yaml
index fc1ca9746..bc315d0cd 100644
--- a/.github/configs/nvidia-master.yaml
+++ b/.github/configs/nvidia-master.yaml
@@ -1737,7 +1737,7 @@ dsv4-fp4-b200-sglang:
       - { tp: 8, ep: 8, dp-attn: true, conc-start: 256, conc-end: 512 }
 
 dsv4-fp4-b200-vllm:
-  image: vllm/vllm-openai:v0.20.0-cu130
+  image: vllm/vllm-openai:v0.21.0
   model: deepseek-ai/DeepSeek-V4-Pro
   model-prefix: dsv4
   runner: b200-dsv4
@@ -1825,7 +1825,7 @@ dsv4-fp4-b200-trt-mtp:
 # MTP variant of dsv4-fp4-b200-vllm. Mirrors the base search space and adds
 # --speculative-config '{"method":"mtp","num_speculative_tokens":2}'.
 dsv4-fp4-b200-vllm-mtp:
-  image: vllm/vllm-openai:v0.20.0-cu130
+  image: vllm/vllm-openai:v0.21.0
   model: deepseek-ai/DeepSeek-V4-Pro
   model-prefix: dsv4
   runner: b200-dsv4
diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index 782b3db37..65d35eeee 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -2769,3 +2769,10 @@
   description:
     - "Update TensorRT-LLM image from v1.1.0rc2.post2 (154d/124d old) to v1.3.0rc14 (latest pre-release)"
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1487
+
+- config-keys:
+    - dsv4-fp4-b200-vllm
+    - dsv4-fp4-b200-vllm-mtp
+  description:
+    - "Update vLLM image from v0.20.0-cu130 (20d/18d old) to v0.21.0"
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1476