Round score
0.7928
+79.28%
Prompts scored
25 / 32
median_e2e_speedup
Below tolerance
7
answered too little to score
Zeroed total
7
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 | +71.18% | 0.592s | 0.171s | 42 | — |
| hf-001 | — | — | — | 17 | candidate output below tolerance |
| hf-003 | +81.88% | 0.751s | 0.136s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 53.30ms | 0.171s | 42 | 15.07ms | — |
| hf-001 | 33.75ms | 0.106s | 17 | 14.99ms | — |
| hf-002 | 46.07ms | 0.163s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 41 |
| — |
| hf-002 | — | — | — | 51 | candidate output below tolerance |
| hf-005 | +83.06% | 0.579s | 0.098s | 31 | — |
| hf-004 | — | — | — | 23 | candidate output below tolerance |
| hf-006 | +82.78% | 0.965s | 0.166s | 57 | — |
| hf-008 | — | — | — | 28 | candidate output below tolerance |
| hf-007 | +82.23% | 0.555s | 0.099s | 32 | — |
| hf-009 | +83.13% | 0.794s | 0.134s | 48 | — |
| hf-010 | +80.19% | 0.951s | 0.188s | 53 | — |
| hf-011 | +86.01% | 0.771s | 0.108s | 42 | — |
| hf-012 | — | — | — | 28 | candidate output below tolerance |
| hf-014 | +78.63% | 0.730s | 0.156s | 42 | — |
| hf-016 | +70.55% | 0.341s | 0.100s | 16 | — |
| hf-018 | +75.21% | 0.401s | 0.100s | 26 | — |
| hf-017 | — | — | — | 23 | candidate output below tolerance |
| hf-015 | +82.70% | 1.146s | 0.198s | 70 | — |
| hf-019 | +77.33% | 0.716s | 0.162s | 41 | — |
| hf-020 | +81.30% | 0.772s | 0.144s | 46 | — |
| hf-013 | +81.60% | 2.259s | 0.416s | 142 | — |
| hf-021 | +78.86% | 0.703s | 0.149s | 42 | — |
| hf-022 | +81.82% | 0.740s | 0.135s | 43 | — |
| hf-023 | — | — | — | 22 | candidate output below tolerance |
| hf-024 | +71.47% | 0.676s | 0.193s | 42 | — |
| hf-025 | +83.74% | 0.762s | 0.124s | 45 | — |
| hf-026 | +83.61% | 0.722s | 0.118s | 43 | — |
| hf-028 | +76.34% | 0.763s | 0.181s | 43 | — |
| hf-027 | +81.39% | 1.006s | 0.187s | 62 | — |
| hf-029 | +81.41% | 0.731s | 0.136s | 46 | — |
| hf-030 | +79.70% | 0.657s | 0.133s | 46 | — |
| hf-031 | +78.30% | 0.667s | 0.145s | 54 | — |
| 51 |
| 14.78ms |
| — |
| hf-003 | 33.67ms | 0.136s | 41 | 15.33ms | — |
| hf-004 | 42.22ms | 0.086s | 23 | 14.87ms | — |
| hf-005 | 39.10ms | 0.098s | 31 | 15.86ms | — |
| hf-006 | 34.10ms | 0.166s | 57 | 15.07ms | — |
| hf-007 | 40.09ms | 0.099s | 32 | 14.74ms | — |
| hf-008 | 32.88ms | 0.106s | 28 | 14.91ms | — |
| hf-009 | 45.76ms | 0.134s | 48 | 15.17ms | — |
| hf-010 | 42.08ms | 0.188s | 53 | 14.96ms | — |
| hf-011 | 35.17ms | 0.108s | 42 | 14.68ms | — |
| hf-012 | 33.46ms | 0.136s | 28 | 15.28ms | — |
| hf-013 | 34.35ms | 0.416s | 142 | 47.21ms | — |
| hf-014 | 63.56ms | 0.156s | 42 | 15.90ms | — |
| hf-015 | 65.57ms | 0.198s | 70 | 15.23ms | — |
| hf-016 | 56.53ms | 0.175s | 45 | 15.54ms | — |
| hf-017 | 39.81ms | 0.098s | 23 | 14.77ms | — |
| hf-018 | 40.41ms | 0.114s | 28 | 15.23ms | — |
| hf-019 | 31.52ms | 0.162s | 46 | 14.80ms | — |
| hf-020 | 41.53ms | 0.159s | 48 | 15.02ms | — |
| hf-021 | 46.53ms | 0.149s | 42 | 15.11ms | — |
| hf-022 | 33.35ms | 0.135s | 43 | 15.52ms | — |
| hf-023 | 33.57ms | 0.150s | 22 | 14.89ms | — |
| hf-024 | 45.95ms | 0.193s | 42 | 15.36ms | — |
| hf-025 | 37.09ms | 0.124s | 45 | 15.67ms | — |
| hf-026 | 31.29ms | 0.118s | 43 | 15.07ms | — |
| hf-027 | 41.77ms | 0.187s | 62 | 14.73ms | — |
| hf-028 | 47.08ms | 0.181s | 44 | 16.18ms | — |
| hf-029 | 45.72ms | 0.136s | 46 | 15.75ms | — |
| hf-030 | 32.25ms | 0.133s | 46 | 14.92ms | — |
| hf-031 | 42.68ms | 0.174s | 68 | 15.44ms | — |
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