how to set online model in vllm benchmark serving?

[liyuerich]

do not want to download model to local path, how to set up --model with online model name on remote server?

python3 ~/vllm/benchmarks/benchmark_serving.py --backend vllm
--model ~/deepseek-R1 --port 8000
--dataset-name random
--random-input 1234
--random-output 2345
--random-range-ratio 0.8
--dataset-path ~/ShareGPT_V3_unfiltered_cleaned_split.json
--max-concurrency 16
--num-prompts 64