Home /
llm.inference.chatbot-short /
019e3b2f-ef77-7467-b840-7fde4b6c308f
meta-llama/Llama-3.1-8B-Instruct
vllm 0.21.0 on 8x NVIDIA H100 80GB HBM3
(fp16) · seed 42 · run 019e3b2f-ef77-7467-b840-7fde4b6c308f
· signed
Headline metrics
TTFT P50 (ms)
25.39
TTFT P99 (ms)
1,678
Throughput (tok/s)
228
$/M tokens
—
J/token
3.26
Power avg (W)
758
Power peak (W)
943
WER mean
—
J / audio s
—
All metrics
compliance_rate
0.9737
cost_source
registry:groq
cost_usd_per_million_tokens
0.0575
energy_joules_total
15,250
joules_per_token
3.26
ok_rate
1.00
power_avg_w
758
power_peak_w
943
req_per_s_all
1.85
req_per_s_passing
1.80
slo_hardware_class
h100
slo_template_resolved
ttft<200ms, tpot<50ms, total<3000ms
throughput_tok_per_s
228
total_p50_ms
863
total_p99_ms
3,021
tpot_p50_ms
6.60
tpot_p99_ms
10.89
ttft_p50_ms
25.39
ttft_p99_ms
1,678
Provenance
Model revision unknown00
Provider vllm
Hardware fingerprint 550474fc9132129654f5d20c316eaffec99a1f67c0cc0b4641e60efb46ea79e7
Dataset builtin-chatbot-short · 6f4b1f68fc3a813baa983cbe70cd9ef57f8c86e6b2e6ccc9aaa2a498e588d510
Timestamp 2026-05-18T13:04:17.783238+00:00
Envelope JSON 49e160db9a2a.json
Verify this result
bench verify /bench/leaderboard/envelopes/49e160db9a2a.json
bench verify re-downloads this envelope, recomputes the canonical
content hash, and validates the Sigstore signature against the embedded
certificate + Rekor entry.