Round score
0.356
+35.60%
Prompts scored
23 / 32
median_e2e_speedup
Below tolerance
9
answered too little to score
Zeroed total
9
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-001 | +42.51% | 0.396s | 0.228s | 18 | — |
| hf-000 | +50.41% | 0.740s | 0.367s | 40 | — |
| hf-002 | +26.61% | 0.528s | 0.388s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 153.88ms | 0.367s | 47 | 84.18ms | — |
| hf-001 | 143.86ms | 0.228s | 18 | 24.01ms | — |
| hf-002 | 173.30ms | 0.428s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 28 |
| — |
| hf-003 | — | — | — | 32 | candidate output below tolerance |
| hf-004 | +25.09% | 0.765s | 0.573s | 41 | — |
| hf-005 | — | — | — | 27 | candidate output below tolerance |
| hf-006 | — | — | — | 35 | candidate output below tolerance |
| hf-007 | +46.88% | 0.778s | 0.413s | 41 | — |
| hf-008 | +47.30% | 0.865s | 0.456s | 42 | — |
| hf-009 | +55.88% | 0.775s | 0.342s | 41 | — |
| hf-010 | — | — | — | 24 | candidate output below tolerance |
| hf-011 | +51.47% | 0.768s | 0.373s | 46 | — |
| hf-012 | +54.04% | 0.565s | 0.260s | 34 | — |
| hf-013 | +50.53% | 0.739s | 0.365s | 40 | — |
| hf-014 | +32.21% | 0.723s | 0.490s | 42 | — |
| hf-015 | +28.17% | 0.821s | 0.590s | 47 | — |
| hf-017 | -20.78% | 0.486s | 0.587s | 28 | — |
| hf-016 | +52.45% | 0.910s | 0.433s | 51 | — |
| hf-018 | — | — | — | 48 | candidate output below tolerance |
| hf-019 | +52.99% | 1.002s | 0.471s | 50 | — |
| hf-020 | +54.07% | 0.941s | 0.432s | 40 | — |
| hf-022 | +42.24% | 0.612s | 0.353s | 24 | — |
| hf-023 | +59.44% | 1.017s | 0.412s | 46 | — |
| hf-021 | +58.68% | 1.020s | 0.422s | 45 | — |
| hf-024 | +32.82% | 1.017s | 0.683s | 47 | — |
| hf-025 | +38.37% | 1.015s | 0.625s | 46 | — |
| hf-026 | — | — | — | 18 | candidate output below tolerance |
| hf-027 | — | — | — | 41 | candidate output below tolerance |
| hf-028 | -7.61% | 0.808s | 0.870s | 41 | — |
| hf-030 | — | — | — | 32 | candidate output below tolerance |
| hf-029 | — | — | — | 22 | candidate output below tolerance |
| hf-031 | +50.77% | 0.987s | 0.486s | 75 | — |
| 43 |
| 82.37ms |
| — |
| hf-003 | 151.51ms | 0.373s | 32 | 105.23ms | — |
| hf-004 | 172.36ms | 0.573s | 41 | 129.18ms | — |
| hf-005 | 231.13ms | 0.578s | 27 | 92.03ms | — |
| hf-006 | 151.76ms | 0.590s | 35 | 114.09ms | — |
| hf-007 | 153.13ms | 0.413s | 41 | 113.82ms | — |
| hf-008 | 187.51ms | 0.542s | 47 | 86.09ms | — |
| hf-009 | 140.98ms | 0.366s | 43 | 86.06ms | — |
| hf-010 | 146.77ms | 0.368s | 24 | 86.92ms | — |
| hf-011 | 157.22ms | 0.373s | 46 | 85.49ms | — |
| hf-012 | 148.92ms | 0.349s | 43 | 88.73ms | — |
| hf-013 | 148.69ms | 0.365s | 42 | 87.59ms | — |
| hf-014 | 140.20ms | 0.490s | 42 | 138.31ms | — |
| hf-015 | 238.56ms | 0.590s | 47 | 92.98ms | — |
| hf-016 | 182.61ms | 0.433s | 51 | 88.52ms | — |
| hf-017 | 161.25ms | 0.643s | 43 | 137.77ms | — |
| hf-018 | 244.54ms | 0.671s | 48 | 107.40ms | — |
| hf-019 | 184.17ms | 0.471s | 50 | 107.48ms | — |
| hf-020 | 180.88ms | 0.432s | 44 | 102.52ms | — |
| hf-021 | 175.28ms | 0.422s | 45 | 85.04ms | — |
| hf-022 | 155.64ms | 0.416s | 52 | 68.89ms | — |
| hf-023 | 130.60ms | 0.633s | 69 | 134.32ms | — |
| hf-024 | 216.55ms | 0.683s | 47 | 133.55ms | — |
| hf-025 | 229.19ms | 0.625s | 46 | 126.36ms | — |
| hf-026 | 166.66ms | 0.274s | 18 | 59.34ms | — |
| hf-027 | 205.33ms | 0.773s | 41 | 190.90ms | — |
| hf-028 | 201.38ms | 0.870s | 42 | 194.97ms | — |
| hf-029 | 380.16ms | 0.717s | 22 | 122.75ms | — |
| hf-030 | 302.63ms | 0.576s | 32 | 87.81ms | — |
| hf-031 | 240.22ms | 0.486s | 75 | 26.82ms | — |
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