Home /
llm.inference.chatbot-short /
019e3b45-8552-7705-ab62-1cb802fe5680
mistralai/Mistral-7B-Instruct-v0.3
vllm 0.21.0 on 8x NVIDIA H100 80GB HBM3
(fp16) · seed 42 · run 019e3b45-8552-7705-ab62-1cb802fe5680
· signed
Headline metrics
TTFT P50 (ms)
20.77
TTFT P99 (ms)
87.92
Throughput (tok/s)
472
$/M tokens
—
J/token
1.88
Power avg (W)
901
Power peak (W)
937
WER mean
—
J / audio s
—
All metrics
compliance_rate
0.9890
energy_joules_total
18,637
joules_per_token
1.88
ok_rate
1.00
power_avg_w
901
power_peak_w
937
req_per_s_all
4.34
req_per_s_passing
4.29
slo_hardware_class
h100
slo_template_resolved
ttft<200ms, tpot<50ms, total<3000ms
throughput_tok_per_s
472
total_p50_ms
817
total_p99_ms
1,096
tpot_p50_ms
7.04
tpot_p99_ms
8.69
ttft_p50_ms
20.77
ttft_p99_ms
87.92
Provenance
Model revision unknown00
Provider vllm
Hardware fingerprint 550474fc9132129654f5d20c316eaffec99a1f67c0cc0b4641e60efb46ea79e7
Dataset builtin-chatbot-short · 6f4b1f68fc3a813baa983cbe70cd9ef57f8c86e6b2e6ccc9aaa2a498e588d510
Timestamp 2026-05-18T13:27:52.402452+00:00
Envelope JSON 010e9504589b.json
Verify this result
bench verify /bench/leaderboard/envelopes/010e9504589b.json
bench verify re-downloads this envelope, recomputes the canonical
content hash, and validates the Sigstore signature against the embedded
certificate + Rekor entry.