Round score
0.6978
+69.78%
Prompts scored
28 / 32
median_e2e_speedup
Below tolerance
4
answered too little to score
Zeroed total
4
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 | +73.84% | 0.758s | 0.198s | 42 | — |
| hf-001 | +72.03% | 0.720s | 0.201s | 41 | — |
| hf-002 | +69.72% | 0.518s | 0.157s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 74.87ms | 0.198s | 42 | 16.14ms | — |
| hf-001 | 77.98ms | 0.201s | 41 | 16.41ms | — |
| hf-002 | 79.22ms | 0.202s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 28 |
| — |
| hf-003 | +72.48% | 0.699s | 0.192s | 47 | — |
| hf-004 | +69.33% | 0.579s | 0.178s | 43 | — |
| hf-005 | — | — | — | 27 | candidate output below tolerance |
| hf-006 | +71.62% | 0.662s | 0.188s | 40 | — |
| hf-007 | +75.51% | 0.711s | 0.174s | 42 | — |
| hf-008 | +54.36% | 0.754s | 0.344s | 46 | — |
| hf-010 | +62.47% | 0.506s | 0.190s | 29 | — |
| hf-009 | +65.44% | 1.099s | 0.380s | 71 | — |
| hf-011 | +67.99% | 0.776s | 0.248s | 51 | — |
| hf-012 | +66.41% | 0.729s | 0.245s | 47 | — |
| hf-013 | +73.10% | 0.711s | 0.191s | 42 | — |
| hf-014 | — | — | — | 28 | candidate output below tolerance |
| hf-015 | +76.04% | 0.737s | 0.177s | 42 | — |
| hf-016 | — | — | — | 28 | candidate output below tolerance |
| hf-017 | +75.58% | 0.725s | 0.177s | 40 | — |
| hf-018 | +72.87% | 0.742s | 0.201s | 42 | — |
| hf-019 | +76.16% | 0.795s | 0.190s | 45 | — |
| hf-020 | +75.36% | 0.760s | 0.187s | 43 | — |
| hf-021 | +74.96% | 0.761s | 0.191s | 43 | — |
| hf-022 | +67.80% | 0.409s | 0.132s | 23 | — |
| hf-023 | +46.88% | 0.685s | 0.364s | 40 | — |
| hf-024 | +72.17% | 0.712s | 0.198s | 42 | — |
| hf-025 | +69.85% | 0.710s | 0.214s | 42 | — |
| hf-027 | +71.34% | 0.500s | 0.143s | 25 | — |
| hf-026 | +72.51% | 0.726s | 0.200s | 43 | — |
| hf-030 | +55.50% | 0.369s | 0.164s | 23 | — |
| hf-028 | — | — | — | 42 | candidate output below tolerance |
| hf-029 | +54.70% | 0.670s | 0.303s | 45 | — |
| hf-031 | +59.95% | 0.367s | 0.147s | 26 | — |
| 44 |
| 16.19ms |
| — |
| hf-003 | 69.33ms | 0.192s | 47 | 16.16ms | — |
| hf-004 | 69.05ms | 0.178s | 43 | 16.11ms | — |
| hf-005 | 66.00ms | 0.174s | 27 | 16.47ms | — |
| hf-006 | 64.90ms | 0.188s | 40 | 16.01ms | — |
| hf-007 | 66.34ms | 0.174s | 42 | 15.94ms | — |
| hf-008 | 95.39ms | 0.375s | 61 | 40.23ms | — |
| hf-009 | 82.75ms | 0.380s | 71 | 105.06ms | — |
| hf-010 | 110.54ms | 0.247s | 42 | 41.86ms | — |
| hf-011 | 81.63ms | 0.248s | 51 | 43.59ms | — |
| hf-012 | 82.19ms | 0.245s | 49 | 39.65ms | — |
| hf-013 | 82.81ms | 0.191s | 42 | 15.94ms | — |
| hf-014 | 69.33ms | 0.162s | 28 | 16.57ms | — |
| hf-015 | 68.68ms | 0.177s | 42 | 16.41ms | — |
| hf-016 | 77.68ms | 0.186s | 28 | 15.99ms | — |
| hf-017 | 69.09ms | 0.177s | 40 | 17.61ms | — |
| hf-018 | 77.04ms | 0.201s | 42 | 16.01ms | — |
| hf-019 | 66.78ms | 0.190s | 45 | 16.39ms | — |
| hf-020 | 78.72ms | 0.187s | 43 | 16.30ms | — |
| hf-021 | 67.19ms | 0.191s | 43 | 16.32ms | — |
| hf-022 | 69.11ms | 0.192s | 45 | 16.11ms | — |
| hf-023 | 95.61ms | 0.615s | 103 | 40.46ms | — |
| hf-024 | 87.83ms | 0.198s | 42 | 16.34ms | — |
| hf-025 | 88.55ms | 0.253s | 48 | 38.66ms | — |
| hf-026 | 92.58ms | 0.200s | 43 | 16.30ms | — |
| hf-027 | 66.36ms | 0.143s | 25 | 16.67ms | — |
| hf-028 | 85.63ms | 0.209s | 42 | 16.25ms | — |
| hf-029 | 72.32ms | 0.303s | 47 | 41.58ms | — |
| hf-030 | 87.49ms | 0.164s | 23 | 15.86ms | — |
| hf-031 | 55.57ms | 0.147s | 26 | 15.65ms | — |
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