| 1 | # Long CLI flags of the Python `vllm bench serve` parser |
| 2 | # (vllm.benchmarks.serve.add_cli_args), sorted; --help excluded. |
| 3 | # Regenerate: .venv/bin/python tests/benchmarks/test_rust_bench_cli_parity.py |
| 4 | # Consumed by rust/src/bench/tests/cli_parity.rs, which requires every flag |
| 5 | # below to be a known vllm-bench flag/alias or a PYTHON_ONLY allowlist entry. |
| 6 | --append-result |
| 7 | --asr-max-audio-len-sec |
| 8 | --asr-min-audio-len-sec |
| 9 | --backend |
| 10 | --base-url |
| 11 | --bfcl-categories |
| 12 | --blazedit-max-distance |
| 13 | --blazedit-min-distance |
| 14 | --burstiness |
| 15 | --chat-template-kwargs |
| 16 | --custom-ensure-client-side-data |
| 17 | --custom-output-len |
| 18 | --dataset-name |
| 19 | --dataset-path |
| 20 | --disable-shuffle |
| 21 | --disable-tqdm |
| 22 | --enable-multimodal-chat |
| 23 | --endpoint |
| 24 | --extra-body |
| 25 | --frequency-penalty |
| 26 | --goodput |
| 27 | --header |
| 28 | --hf-name |
| 29 | --hf-output-len |
| 30 | --hf-split |
| 31 | --hf-subset |
| 32 | --host |
| 33 | --ignore-eos |
| 34 | --input-len |
| 35 | --insecure |
| 36 | --label |
| 37 | --logprobs |
| 38 | --lora-assignment |
| 39 | --lora-modules |
| 40 | --max-concurrency |
| 41 | --metadata |
| 42 | --metric-percentiles |
| 43 | --min-p |
| 44 | --model |
| 45 | --no-oversample |
| 46 | --no-reranker |
| 47 | --no-self-timed |
| 48 | --no-stream |
| 49 | --num-prompts |
| 50 | --num-warmups |
| 51 | --output-len |
| 52 | --percentile-metrics |
| 53 | --plot-dataset-stats |
| 54 | --plot-timeline |
| 55 | --port |
| 56 | --prefix-repetition-num-prefixes |
| 57 | --prefix-repetition-output-len |
| 58 | --prefix-repetition-prefix-len |
| 59 | --prefix-repetition-suffix-len |
| 60 | --presence-penalty |
| 61 | --probe-request-rate |
| 62 | --profile |
| 63 | --ramp-up-end-rps |
| 64 | --ramp-up-start-rps |
| 65 | --ramp-up-strategy |
| 66 | --random-batch-size |
| 67 | --random-input-len |
| 68 | --random-mm-base-items-per-request |
| 69 | --random-mm-bucket-config |
| 70 | --random-mm-limit-mm-per-prompt |
| 71 | --random-mm-num-mm-items-range-ratio |
| 72 | --random-output-len |
| 73 | --random-prefix-len |
| 74 | --random-range-ratio |
| 75 | --ready-check-timeout-sec |
| 76 | --repetition-penalty |
| 77 | --request-id-prefix |
| 78 | --request-rate |
| 79 | --result-dir |
| 80 | --result-filename |
| 81 | --save-detailed |
| 82 | --save-result |
| 83 | --seed |
| 84 | --self-timed |
| 85 | --served-model-name |
| 86 | --sharegpt-output-len |
| 87 | --skip-chat-template |
| 88 | --skip-tokenizer-init |
| 89 | --sonnet-input-len |
| 90 | --sonnet-output-len |
| 91 | --sonnet-prefix-len |
| 92 | --spec-bench-category |
| 93 | --spec-bench-output-len |
| 94 | --speed-bench-category |
| 95 | --speed-bench-dataset-subset |
| 96 | --speed-bench-output-len |
| 97 | --temperature |
| 98 | --timed-trace-chunk-hash-size |
| 99 | --timed-trace-label-hash-ids |
| 100 | --timed-trace-label-input-length |
| 101 | --timed-trace-label-output-length |
| 102 | --timed-trace-label-timestamp |
| 103 | --timed-trace-sec-multiplier |
| 104 | --timeline-itl-thresholds |
| 105 | --tokenizer |
| 106 | --tokenizer-mode |
| 107 | --top-k |
| 108 | --top-p |
| 109 | --trust-remote-code |
| 110 | --use-beam-search |