InferenceBench

Qwen/Qwen2.5-7B-Instruct

vllm unknown on 8x NVIDIA H100 80GB HBM3 (bf16) · seed 0 · run 019ebfa5-10fc-7a12-97b0-bd7ec6dc75be · signed

Headline metrics

TTFT P50 (ms) 12.99
TTFT P99 (ms)
Throughput (tok/s)
$/M tokens
J/token 21.40
Power avg (W) 643
Power peak (W) 848
WER mean
J / audio s

All metrics

chrf_mean 0.8912
chrf_p50 0.9439
chrf_p95 1.00
energy_joules_total 2,439
joules_per_token 21.40
n_ok 8.00
n_samples 8.00
ok_rate 1.00
power_avg_w 643
power_peak_w 848
tokens_out_total 114
total_p50_ms 114
ttft_p50_ms 12.99

Provenance

Model revisionunknown00
Providervllm
Hardware fingerprintc353425077b20b36c71d9c32d9e5d49096478f4a0f7d6078c1dfc83c408ccf50
Datasetbuiltin-flores-mini-en-fr · 25cb8c32504c769c88e3b41ad01bb304ccd100dbc9278b3a41f520053512185e
Timestamp2026-06-13T06:22:06.588570+00:00
Envelope JSON4f132764ce45.json

Verify this result

bench verify /bench/leaderboard/envelopes/4f132764ce45.json

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