Round score
0.6545
+65.45%
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-001 | +71.49% | 0.700s | 0.199s | 42 | — |
| hf-002 | +74.58% | 0.657s | 0.167s | 37 | — |
| hf-000 | — | — | — |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 98.66ms | 0.340s | 46 | 40.99ms | — |
| hf-001 | 80.94ms | 0.199s | 42 | 15.29ms | — |
| hf-002 | 63.72ms | 0.196s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 46 |
| candidate output below tolerance |
| hf-003 | +73.41% | 0.710s | 0.189s | 43 | — |
| hf-005 | +62.40% | 0.374s | 0.141s | 24 | — |
| hf-004 | +75.68% | 0.773s | 0.188s | 48 | — |
| hf-007 | +55.70% | 0.372s | 0.165s | 22 | — |
| hf-006 | +60.46% | 0.716s | 0.283s | 43 | — |
| hf-008 | +62.47% | 0.860s | 0.323s | 53 | — |
| hf-009 | +71.08% | 0.726s | 0.210s | 43 | — |
| hf-010 | +60.38% | 0.724s | 0.287s | 43 | — |
| hf-011 | +70.13% | 0.683s | 0.204s | 43 | — |
| hf-012 | +53.64% | 0.591s | 0.274s | 41 | — |
| hf-013 | +70.53% | 0.716s | 0.211s | 42 | — |
| hf-014 | +60.35% | 0.705s | 0.279s | 41 | — |
| hf-015 | +61.69% | 0.736s | 0.282s | 41 | — |
| hf-016 | +73.13% | 0.704s | 0.189s | 42 | — |
| hf-018 | +70.18% | 0.693s | 0.207s | 43 | — |
| hf-017 | +64.74% | 0.947s | 0.334s | 61 | — |
| hf-019 | +67.66% | 0.612s | 0.198s | 42 | — |
| hf-021 | +66.25% | 0.460s | 0.155s | 28 | — |
| hf-020 | — | — | — | 20 | candidate output below tolerance |
| hf-022 | +60.80% | 0.367s | 0.144s | 23 | — |
| hf-024 | +66.53% | 0.414s | 0.139s | 28 | — |
| hf-023 | — | — | — | 42 | candidate output below tolerance |
| hf-026 | +65.83% | 0.529s | 0.181s | 28 | — |
| hf-025 | +74.53% | 0.784s | 0.200s | 46 | — |
| hf-027 | +70.61% | 0.716s | 0.210s | 43 | — |
| hf-029 | +30.53% | 0.302s | 0.209s | 17 | — |
| hf-028 | +64.52% | 0.612s | 0.217s | 40 | — |
| hf-030 | +66.96% | 0.615s | 0.203s | 47 | — |
| hf-031 | +65.07% | 0.762s | 0.266s | 68 | — |
| 45 |
| 16.30ms |
| — |
| hf-003 | 69.69ms | 0.189s | 43 | 16.57ms | — |
| hf-004 | 55.83ms | 0.188s | 48 | 16.22ms | — |
| hf-005 | 65.56ms | 0.213s | 49 | 15.96ms | — |
| hf-006 | 111.69ms | 0.283s | 43 | 67.10ms | — |
| hf-007 | 104.75ms | 0.208s | 43 | 16.44ms | — |
| hf-008 | 79.22ms | 0.323s | 53 | 40.73ms | — |
| hf-009 | 91.66ms | 0.277s | 46 | 67.23ms | — |
| hf-010 | 107.73ms | 0.287s | 43 | 38.08ms | — |
| hf-011 | 86.88ms | 0.204s | 43 | 15.19ms | — |
| hf-012 | 115.29ms | 0.274s | 41 | 67.41ms | — |
| hf-013 | 92.88ms | 0.211s | 42 | 15.94ms | — |
| hf-014 | 120.24ms | 0.279s | 41 | 67.79ms | — |
| hf-015 | 99.35ms | 0.298s | 46 | 39.61ms | — |
| hf-016 | 86.33ms | 0.189s | 42 | 15.97ms | — |
| hf-017 | 92.82ms | 0.334s | 61 | 40.24ms | — |
| hf-018 | 87.35ms | 0.207s | 45 | 15.95ms | — |
| hf-019 | 78.64ms | 0.198s | 42 | 16.99ms | — |
| hf-020 | 93.24ms | 0.183s | 20 | 15.78ms | — |
| hf-021 | 65.61ms | 0.155s | 28 | 15.98ms | — |
| hf-022 | 55.36ms | 0.144s | 23 | 16.18ms | — |
| hf-023 | 64.41ms | 0.168s | 42 | 16.03ms | — |
| hf-024 | 63.83ms | 0.167s | 42 | 16.40ms | — |
| hf-025 | 67.00ms | 0.200s | 46 | 15.80ms | — |
| hf-026 | 77.25ms | 0.181s | 28 | 16.24ms | — |
| hf-027 | 77.53ms | 0.277s | 47 | 66.05ms | — |
| hf-028 | 98.95ms | 0.289s | 45 | 71.67ms | — |
| hf-029 | 133.38ms | 0.209s | 17 | 16.25ms | — |
| hf-030 | 70.83ms | 0.203s | 47 | 16.01ms | — |
| hf-031 | 76.22ms | 0.266s | 68 | 15.75ms | — |
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