diff --git a/.github/configs/amd-master.yaml b/.github/configs/amd-master.yaml
index 3ec6b95c9..de7f5e62a 100644
--- a/.github/configs/amd-master.yaml
+++ b/.github/configs/amd-master.yaml
@@ -495,6 +495,28 @@ minimaxm2.5-fp8-mi355x-atom:
     - { tp: 4, conc-start: 4, conc-end: 128 }
     - { tp: 8, ep: 8, conc-start: 32, conc-end: 256 }
 
+minimaxm2.5-fp4-mi355x-vllm:
+  image: vllm/vllm-openai-rocm:v0.19.1
+  model: amd/MiniMax-M2.5-MXFP4
+  model-prefix: minimaxm2.5
+  runner: mi355x
+  precision: fp4
+  framework: vllm
+  multinode: false
+  seq-len-configs:
+  - isl: 1024
+    osl: 1024
+    search-space:
+    - { tp: 1, conc-start: 4, conc-end: 32 }
+    - { tp: 2, conc-start: 4, conc-end: 64 }
+    - { tp: 4, conc-start: 4, conc-end: 64 }
+  - isl: 8192
+    osl: 1024
+    search-space:
+    - { tp: 1, conc-start: 4, conc-end: 32 }
+    - { tp: 2, conc-start: 4, conc-end: 64 }
+    - { tp: 4, conc-start: 4, conc-end: 64 }
+
 minimaxm2.5-fp8-mi300x-vllm:
   image: vllm/vllm-openai-rocm:v0.16.0
   model: MiniMaxAI/MiniMax-M2.5
diff --git a/benchmarks/single_node/minimaxm2.5_fp4_mi355x.sh b/benchmarks/single_node/minimaxm2.5_fp4_mi355x.sh
new file mode 100755
index 000000000..116a1c325
--- /dev/null
+++ b/benchmarks/single_node/minimaxm2.5_fp4_mi355x.sh
@@ -0,0 +1,94 @@
+#!/usr/bin/env bash
+
+source "$(dirname "$0")/../benchmark_lib.sh"
+
+check_env_vars \
+    MODEL \
+    TP \
+    EP_SIZE \
+    CONC \
+    ISL \
+    OSL \
+    MAX_MODEL_LEN \
+    RANDOM_RANGE_RATIO \
+    RESULT_FILENAME
+
+if [[ -n "$SLURM_JOB_ID" ]]; then
+  echo "JOB $SLURM_JOB_ID running on $SLURMD_NODENAME"
+fi
+
+hf download "$MODEL"
+
+# Set HIP_VISIBLE_DEVICES to match ROCR_VISIBLE_DEVICES for Ray compatibility in vLLM 0.14+
+if [ -n "$ROCR_VISIBLE_DEVICES" ]; then
+    export HIP_VISIBLE_DEVICES="$ROCR_VISIBLE_DEVICES"
+fi
+
+export VLLM_ROCM_USE_AITER=1
+EXTRA_VLLM_ARGS=""
+# if [ "$TP" -ge 4 ]; then
+#     # AITER CK fused MoE kernels lack compiled tiles for N=intermediate_size/TP
+#     # when TP>=4 (TP=4, N=384). Disable AITER MoE to fall back to triton, but keep
+#     # AITER attention. See: https://github.com/vllm-project/vllm/issues/35637
+#     export VLLM_ROCM_USE_AITER_MOE=0
+#     EXTRA_VLLM_ARGS="--attention-backend ROCM_AITER_UNIFIED_ATTN"
+#     pip install amd-quark 2>/dev/null || true
+# fi
+
+SERVER_LOG=/workspace/server.log
+PORT=${PORT:-8888}
+
+if [ "${EVAL_ONLY}" = "true" ]; then
+    setup_eval_context
+    MAX_MODEL_LEN="$EVAL_MAX_MODEL_LEN"
+fi
+
+if [ "$EP_SIZE" -gt 1 ]; then
+  EP=" --enable-expert-parallel"
+else
+  EP=" "
+fi
+
+# Start GPU monitoring (power, temperature, clocks every second)
+start_gpu_monitor
+
+set -x
+vllm serve $MODEL --port $PORT \
+--tensor-parallel-size=$TP \
+$EP \
+--gpu-memory-utilization 0.95 \
+--max-model-len $MAX_MODEL_LEN \
+--kv-cache-dtype fp8 \
+--block-size=32 \
+--no-enable-prefix-caching \
+--attention-backend "ROCM_AITER_FA" \
+--trust-remote-code \
+$EXTRA_VLLM_ARGS > $SERVER_LOG 2>&1 &
+
+SERVER_PID=$!
+
+# Wait for server to be ready
+wait_for_server_ready --port "$PORT" --server-log "$SERVER_LOG" --server-pid "$SERVER_PID"
+
+run_benchmark_serving \
+    --model "$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 "$RESULT_FILENAME" \
+    --result-dir /workspace/ \
+    --trust-remote-code
+
+# After throughput, run evaluation only if RUN_EVAL is true
+if [ "${RUN_EVAL}" = "true" ]; then
+    run_eval --framework lm-eval --port "$PORT"
+    append_lm_eval_summary
+fi
+
+# Stop GPU monitoring
+stop_gpu_monitor
+set +x
diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index c0d1c1333..6bd6c92e3 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -1282,6 +1282,16 @@
     - "Upgrade vLLM image to v0.19.0"
     - "Enable FP8 KV cache + AITER FA for minimaxm2.5-fp8-mi355x-vllm"
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1003
+  
+- config-keys:
+    - minimaxm2.5-fp4-mi355x-vllm
+  description:
+    - "Add MiniMax M2.5 MXFP4 vLLM benchmark for MI355X"
+    - "Model: amd/MiniMax-M2.5-MXFP4 with --trust-remote-code and --block-size=32"
+    - "Image: vllm/vllm-openai-rocm:v0.19.1"
+    - "Environment: VLLM_ROCM_USE_AITER=1"
+    - "Tp=1, TP=2 and TP=4, concurrency 4-64 for 1k1k, 1k8k, and 8k1k sequence lengths"
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/827
 
 - config-keys:
     - qwen3.5-fp8-h200-sglang-mtp