Round score
0.6195
+61.95%
Prompts scored
26 / 32
median_e2e_speedup
Below tolerance
6
answered too little to score
Zeroed total
6
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 | +75.25% | 0.758s | 0.188s | 42 | — |
| hf-001 | +70.47% | 0.731s | 0.216s | 42 | — |
| hf-002 | +67.31% | 0.518s | 0.169s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 83.14ms | 0.188s | 42 | 17.85ms | — |
| hf-001 | 76.96ms | 0.216s | 42 | 17.74ms | — |
| hf-002 | 99.21ms | 0.186s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 28 |
| — |
| hf-003 | +74.97% | 0.699s | 0.175s | 47 | — |
| hf-004 | +66.67% | 0.579s | 0.193s | 43 | — |
| hf-005 | +61.94% | 0.769s | 0.293s | 49 | — |
| hf-006 | +58.84% | 0.673s | 0.277s | 42 | — |
| hf-007 | +71.86% | 0.711s | 0.200s | 42 | — |
| hf-008 | +54.57% | 0.754s | 0.343s | 46 | — |
| hf-010 | +43.18% | 0.506s | 0.288s | 29 | — |
| hf-009 | +48.24% | 0.977s | 0.506s | 67 | — |
| hf-011 | +61.63% | 0.776s | 0.298s | 51 | — |
| hf-012 | +62.90% | 0.729s | 0.271s | 47 | — |
| hf-013 | +70.03% | 0.711s | 0.213s | 42 | — |
| hf-014 | — | — | — | 22 | candidate output below tolerance |
| hf-015 | +73.71% | 0.737s | 0.194s | 42 | — |
| hf-016 | +58.04% | 0.759s | 0.319s | 43 | — |
| hf-017 | +71.28% | 0.725s | 0.208s | 40 | — |
| hf-018 | +59.87% | 0.731s | 0.293s | 41 | — |
| hf-019 | — | — | — | 18 | candidate output below tolerance |
| hf-020 | — | — | — | 16 | candidate output below tolerance |
| hf-021 | +74.38% | 0.761s | 0.195s | 43 | — |
| hf-022 | +54.25% | 0.409s | 0.187s | 23 | — |
| hf-023 | — | — | — | 15 | candidate output below tolerance |
| hf-024 | +77.01% | 0.712s | 0.164s | 42 | — |
| hf-025 | +72.56% | 0.710s | 0.195s | 42 | — |
| hf-027 | +73.56% | 0.498s | 0.132s | 24 | — |
| hf-026 | +77.06% | 0.726s | 0.167s | 43 | — |
| hf-030 | +61.96% | 0.369s | 0.141s | 23 | — |
| hf-028 | — | — | — | 42 | candidate output below tolerance |
| hf-029 | — | — | — | 18 | candidate output below tolerance |
| hf-031 | +54.72% | 0.367s | 0.166s | 26 | — |
| 44 |
| 18.09ms |
| — |
| hf-003 | 70.94ms | 0.175s | 47 | 18.01ms | — |
| hf-004 | 72.11ms | 0.193s | 43 | 17.92ms | — |
| hf-005 | 77.39ms | 0.293s | 57 | 75.65ms | — |
| hf-006 | 102.13ms | 0.277s | 42 | 70.98ms | — |
| hf-007 | 95.47ms | 0.200s | 42 | 18.27ms | — |
| hf-008 | 119.48ms | 0.360s | 61 | 66.03ms | — |
| hf-009 | 106.16ms | 0.506s | 67 | 95.69ms | — |
| hf-010 | 135.81ms | 0.326s | 43 | 66.51ms | — |
| hf-011 | 105.36ms | 0.316s | 56 | 66.84ms | — |
| hf-012 | 101.15ms | 0.271s | 47 | 64.19ms | — |
| hf-013 | 92.25ms | 0.213s | 42 | 17.68ms | — |
| hf-014 | 96.37ms | 0.201s | 22 | 18.11ms | — |
| hf-015 | 89.39ms | 0.194s | 42 | 17.69ms | — |
| hf-016 | 95.97ms | 0.319s | 43 | 104.77ms | — |
| hf-017 | 126.11ms | 0.208s | 43 | 17.71ms | — |
| hf-018 | 98.34ms | 0.293s | 41 | 74.92ms | — |
| hf-019 | 93.15ms | 0.146s | 18 | 18.16ms | — |
| hf-020 | 92.69ms | 0.123s | 16 | 17.05ms | — |
| hf-021 | 74.04ms | 0.195s | 43 | 17.88ms | — |
| hf-022 | 82.50ms | 0.187s | 23 | 17.91ms | — |
| hf-023 | 101.34ms | 0.137s | 15 | 17.99ms | — |
| hf-024 | 75.78ms | 0.164s | 42 | 18.59ms | — |
| hf-025 | 73.20ms | 0.195s | 42 | 17.80ms | — |
| hf-026 | 79.58ms | 0.167s | 43 | 18.06ms | — |
| hf-027 | 64.53ms | 0.132s | 24 | 17.70ms | — |
| hf-028 | 94.35ms | 0.285s | 42 | 69.83ms | — |
| hf-029 | 92.64ms | 0.145s | 18 | 18.16ms | — |
| hf-030 | 70.86ms | 0.141s | 23 | 18.10ms | — |
| hf-031 | 63.81ms | 0.166s | 26 | 18.28ms | — |
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