InferenceBench

Qwen/Qwen2-VL-7B-Instruct

vllm unknown on 8x NVIDIA H100 80GB HBM3 (fp16) · seed 0 · run 019e3b85-85be-7975-8434-5e6de057cb2b · signed

Headline metrics

TTFT P50 (ms) 84.89
TTFT P99 (ms)
Throughput (tok/s)
$/M tokens
J/token
Power avg (W)
Power peak (W)
WER mean
J / audio s

All metrics

accuracy 0.6000
accuracy_p05 0.0000
accuracy_p50 1.00
accuracy_p95 1.00
n_ok 5.00
n_samples 5.00
ok_rate 1.00
tokens_out_total 13.00
total_p50_ms 84.89
ttft_p50_ms 84.89

Provenance

Model revisionunknown00
Providervllm
Hardware fingerprint550474fc9132129654f5d20c316eaffec99a1f67c0cc0b4641e60efb46ea79e7
Datasetbuiltin-chart-qa-mini · 2057d9889a63d42a0fa66599cb0ca9b9233cb4f656dffe886705b4a03229a1b9
Timestamp2026-05-18T14:37:46.814665+00:00
Envelope JSON7b8d5bcf8a61.json

Verify this result

bench verify /bench/leaderboard/envelopes/7b8d5bcf8a61.json

bench verify re-downloads this envelope, recomputes the canonical content hash, and validates the Sigstore signature against the embedded certificate + Rekor entry.