Round score
0.7145
+71.45%
Prompts scored
29 / 32
median_e2e_speedup
Below tolerance
3
answered too little to score
Zeroed total
3
scored 0.0 without measuring
The round score is the scoring rule applied to the prompts below. Both engines are compared at the same output token count, so a candidate that stops early cannot buy speed by answering less: a prompt under the tolerance bar scores nothing and shows its gate reason.
32 prompts
| Prompt | Speedup | Baseline | Candidate | Tokens | Gate |
|---|---|---|---|---|---|
| hf-000 | +56.20% | 0.591s | 0.259s | 34 | — |
| hf-001 | +74.45% | 0.713s | 0.182s | 41 | — |
| hf-002 | +67.00% | 0.773s | 0.255s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 116.91ms | 0.298s | 46 | 40.92ms | — |
| hf-001 | 98.11ms | 0.182s | 41 | 17.41ms | — |
| hf-002 | 132.73ms | 0.255s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 46 |
| — |
| hf-003 | +70.67% | 0.709s | 0.208s | 43 | — |
| hf-004 | +72.76% | 0.741s | 0.202s | 42 | — |
| hf-005 | +73.16% | 0.715s | 0.192s | 43 | — |
| hf-006 | +65.99% | 0.735s | 0.250s | 46 | — |
| hf-007 | +61.05% | 0.734s | 0.286s | 43 | — |
| hf-008 | +71.39% | 0.730s | 0.209s | 43 | — |
| hf-009 | +72.70% | 0.744s | 0.203s | 40 | — |
| hf-010 | +76.28% | 0.774s | 0.184s | 42 | — |
| hf-011 | +76.16% | 0.774s | 0.185s | 43 | — |
| hf-012 | +67.32% | 0.771s | 0.252s | 44 | — |
| hf-013 | +71.51% | 0.731s | 0.208s | 40 | — |
| hf-015 | +62.12% | 0.376s | 0.142s | 20 | — |
| hf-017 | +65.96% | 0.466s | 0.158s | 25 | — |
| hf-016 | +71.00% | 0.712s | 0.207s | 41 | — |
| hf-014 | — | — | — | 31 | candidate output below tolerance |
| hf-019 | +69.51% | 0.526s | 0.160s | 29 | — |
| hf-018 | +71.80% | 0.779s | 0.220s | 46 | — |
| hf-021 | +76.66% | 0.526s | 0.123s | 25 | — |
| hf-020 | — | — | — | 22 | candidate output below tolerance |
| hf-022 | — | — | — | 24 | candidate output below tolerance |
| hf-023 | +73.79% | 0.755s | 0.198s | 42 | — |
| hf-025 | +73.04% | 0.746s | 0.201s | 42 | — |
| hf-024 | +78.96% | 0.787s | 0.166s | 46 | — |
| hf-026 | +76.61% | 0.765s | 0.179s | 43 | — |
| hf-027 | +72.68% | 0.734s | 0.200s | 39 | — |
| hf-028 | +74.27% | 0.752s | 0.193s | 41 | — |
| hf-029 | +74.88% | 0.680s | 0.171s | 41 | — |
| hf-030 | +52.95% | 0.533s | 0.251s | 34 | — |
| hf-031 | +60.89% | 0.533s | 0.208s | 42 | — |
| 46 |
| 42.72ms |
| — |
| hf-003 | 90.86ms | 0.208s | 43 | 17.18ms | — |
| hf-004 | 83.96ms | 0.218s | 45 | 17.69ms | — |
| hf-005 | 90.54ms | 0.192s | 43 | 17.44ms | — |
| hf-006 | 107.25ms | 0.250s | 46 | 40.78ms | — |
| hf-007 | 89.43ms | 0.286s | 43 | 47.35ms | — |
| hf-008 | 91.43ms | 0.209s | 43 | 17.96ms | — |
| hf-009 | 86.02ms | 0.203s | 40 | 17.64ms | — |
| hf-010 | 83.38ms | 0.200s | 46 | 17.07ms | — |
| hf-011 | 67.37ms | 0.217s | 55 | 17.12ms | — |
| hf-012 | 92.63ms | 0.252s | 44 | 41.66ms | — |
| hf-013 | 107.81ms | 0.208s | 40 | 17.39ms | — |
| hf-014 | 71.81ms | 0.172s | 31 | 17.47ms | — |
| hf-015 | 91.60ms | 0.192s | 46 | 17.07ms | — |
| hf-016 | 88.66ms | 0.250s | 48 | 43.86ms | — |
| hf-017 | 74.66ms | 0.158s | 25 | 17.23ms | — |
| hf-018 | 84.84ms | 0.220s | 46 | 17.56ms | — |
| hf-019 | 92.37ms | 0.160s | 29 | 17.20ms | — |
| hf-020 | 65.82ms | 0.167s | 22 | 18.02ms | — |
| hf-021 | 55.07ms | 0.123s | 25 | 17.32ms | — |
| hf-022 | 65.49ms | 0.149s | 24 | 17.39ms | — |
| hf-023 | 64.32ms | 0.198s | 42 | 17.32ms | — |
| hf-024 | 64.72ms | 0.166s | 46 | 17.29ms | — |
| hf-025 | 66.30ms | 0.201s | 43 | 17.46ms | — |
| hf-026 | 78.31ms | 0.179s | 43 | 17.53ms | — |
| hf-027 | 66.41ms | 0.200s | 39 | 17.62ms | — |
| hf-028 | 76.86ms | 0.193s | 41 | 17.01ms | — |
| hf-029 | 70.46ms | 0.171s | 41 | 17.89ms | — |
| hf-030 | 107.43ms | 0.291s | 46 | 42.10ms | — |
| hf-031 | 90.89ms | 0.225s | 51 | 17.55ms | — |
evidence
requests per s
sla goodput ratio
output tokens per s
p99 e2e ms rel range
p99 itl ms rel range
p99 ttft ms rel range
Image digest
Patch hash