Update README.md

README.md

@@ -346,13 +346,13 @@ Run the benchmarks under `vllm` root folder:
 ### baseline
 ```Shell
 export MODEL=microsoft/Phi-4-mini-instruct
-python benchmarks/benchmark_latency.py --input-len 256 --output-len 256 --model $MODEL --batch-size 1
+vllm bench latency --input-len 256 --output-len 256 --model $MODEL --batch-size 1
 ```
 
 ### AWQ-INT4
 ```Shell
 export MODEL=pytorch/Phi-4-mini-instruct-AWQ-INT4
-VLLM_DISABLE_COMPILE_CACHE=1 python benchmarks/benchmark_latency.py --input-len 256 --output-len 256 --model $MODEL --batch-size 1
+VLLM_DISABLE_COMPILE_CACHE=1 vllm bench latency --input-len 256 --output-len 256 --model $MODEL --batch-size 1
 ```
 </details>