diff --git a/.github/configs/nvidia-master.yaml b/.github/configs/nvidia-master.yaml
index 543864d4a..ee122cca4 100644
--- a/.github/configs/nvidia-master.yaml
+++ b/.github/configs/nvidia-master.yaml
@@ -1804,6 +1804,34 @@ qwen3.5-bf16-b200-sglang:
     search-space:
     - { tp: 8, ep: 1, conc-start: 4, conc-end: 64 }
 
+qwen3.5-fp4-b200-sglang:
+  image: lmsysorg/sglang:v0.5.9-cu129-amd64
+  model: nvidia/Qwen3.5-397B-A17B-NVFP4
+  model-prefix: qwen3.5
+  runner: b200
+  precision: fp4
+  framework: sglang
+  multinode: false
+  seq-len-configs:
+  - isl: 1024
+    osl: 1024
+    search-space:
+    - { tp: 4, ep: 1, conc-start: 4, conc-end: 64 }
+    - { tp: 8, ep: 1, conc-start: 4, conc-end: 4 }
+    - { tp: 8, ep: 8, conc-start: 64, conc-end: 128 }
+  - isl: 1024
+    osl: 8192
+    search-space:
+    - { tp: 4, ep: 1, conc-start: 4, conc-end: 64 }
+    - { tp: 8, ep: 1, conc-start: 4, conc-end: 4 }
+    - { tp: 8, ep: 1, conc-start: 64, conc-end: 128 }
+  - isl: 8192
+    osl: 1024
+    search-space:
+    - { tp: 4, ep: 1, conc-start: 4, conc-end: 64 }
+    - { tp: 8, ep: 1, conc-start: 4, conc-end: 4 }
+    - { tp: 8, ep: 1, conc-start: 64, conc-end: 128 }
+
 kimik2.5-int4-b200-vllm:
   image: vllm/vllm-openai:v0.15.1
   model: moonshotai/Kimi-K2.5
diff --git a/benchmarks/single_node/qwen3.5_fp4_b200.sh b/benchmarks/single_node/qwen3.5_fp4_b200.sh
new file mode 100755
index 000000000..9773e7f58
--- /dev/null
+++ b/benchmarks/single_node/qwen3.5_fp4_b200.sh
@@ -0,0 +1,71 @@
+#!/usr/bin/env bash
+
+source "$(dirname "$0")/../benchmark_lib.sh"
+
+check_env_vars \
+    MODEL \
+    TP \
+    CONC \
+    ISL \
+    OSL \
+    RANDOM_RANGE_RATIO \
+    RESULT_FILENAME \
+    EP_SIZE
+
+if [[ -n "$SLURM_JOB_ID" ]]; then
+  echo "JOB $SLURM_JOB_ID running on $SLURMD_NODENAME"
+fi
+
+nvidia-smi
+
+hf download "$MODEL"
+
+export NCCL_NVLS_ENABLE=1
+export SGL_ENABLE_JIT_DEEPGEMM=false
+export SGLANG_ENABLE_FLASHINFER_GEMM=true
+export PYTHONUNBUFFERED=1
+
+SERVER_LOG=/workspace/server.log
+PORT=${PORT:-8888}
+CONTEXT_LENGTH=$((ISL + OSL + 20))
+CUDA_GRAPH_MAX_BS=$CONC
+MAX_RUNNING_REQUESTS=$((CONC > 128 ? CONC : 128))
+MEM_FRAC_STATIC=0.85
+
+echo "Config: ISL=$ISL, OSL=$OSL, CONC=$CONC, EP=$EP_SIZE, MEM=$MEM_FRAC_STATIC, CUDA_BS=$CUDA_GRAPH_MAX_BS, MAX_RR=$MAX_RUNNING_REQUESTS"
+
+set -x
+PYTHONNOUSERSITE=1 python3 -m sglang.launch_server --model-path=$MODEL --host=0.0.0.0 --port=$PORT \
+--trust-remote-code \
+--tensor-parallel-size=$TP --data-parallel-size=1 --ep-size $EP_SIZE \
+--cuda-graph-max-bs $CUDA_GRAPH_MAX_BS --max-running-requests $MAX_RUNNING_REQUESTS \
+--mem-fraction-static $MEM_FRAC_STATIC --chunked-prefill-size 32768 --max-prefill-tokens 32768 \
+--context-length $CONTEXT_LENGTH --disable-radix-cache \
+--attention-backend trtllm_mha --moe-runner-backend flashinfer_trtllm \
+--enable-flashinfer-allreduce-fusion --scheduler-recv-interval 30 \
+--stream-interval 30 --quantization modelopt_fp4 \
+--kv-cache-dtype fp8_e4m3 --fp4-gemm-backend flashinfer_cutlass > $SERVER_LOG 2>&1 &
+
+SERVER_PID=$!
+
+wait_for_server_ready --port "$PORT" --server-log "$SERVER_LOG" --server-pid "$SERVER_PID"
+
+pip install -q datasets pandas
+
+run_benchmark_serving \
+    --model "$MODEL" \
+    --port "$PORT" \
+    --backend vllm \
+    --input-len "$ISL" \
+    --output-len "$OSL" \
+    --random-range-ratio "$RANDOM_RANGE_RATIO" \
+    --num-prompts "$((CONC * 10))" \
+    --max-concurrency "$CONC" \
+    --result-filename "$RESULT_FILENAME" \
+    --result-dir /workspace/
+
+if [ "${RUN_EVAL}" = "true" ]; then
+    run_eval --framework lm-eval --port "$PORT" --concurrent-requests $CONC
+    append_lm_eval_summary
+fi
+set +x
diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index 006522fc8..0df5b5dad 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -824,3 +824,12 @@
     - "Uses triton attention backend, TP=8, concurrency 4-64"
     - "Following AMD Andy Luo's recipe"
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/850
+
+- config-keys:
+    - qwen3.5-fp4-b200-sglang
+  description:
+    - "Add Qwen3.5-397B-A17B NVFP4 B200 SGLang benchmark config and launch script"
+    - "Image: lmsysorg/sglang:v0.5.9-cu129-amd64"
+    - "Model: nvidia/Qwen3.5-397B-A17B-NVFP4"
+    - "Configs: 1k1k (conc 4-64 ep1, conc 128 ep8), 1k8k (conc 4-128 ep1), 8k1k (conc 4-128 ep1)"
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/820