Round score
0.7235
+72.35%
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 | — | — | — | 42 | candidate output below tolerance |
| hf-002 | +64.42% | 0.726s | 0.258s | 42 | — |
| hf-003 | +72.07% | 0.706s | 0.197s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 99.52ms | 0.196s | 42 | 18.41ms | — |
| hf-001 | 128.81ms | 0.313s | 40 | 53.57ms | — |
| hf-002 | 112.90ms | 0.288s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 40 |
| — |
| hf-004 | +73.26% | 0.740s | 0.198s | 43 | — |
| hf-005 | +77.10% | 0.744s | 0.170s | 40 | — |
| hf-006 | +80.86% | 0.924s | 0.177s | 46 | — |
| hf-007 | +75.98% | 0.610s | 0.147s | 33 | — |
| hf-008 | +65.71% | 0.895s | 0.307s | 46 | — |
| hf-009 | +65.90% | 0.773s | 0.264s | 43 | — |
| hf-001 | — | — | — | 40 | candidate output below tolerance |
| hf-010 | +73.00% | 0.797s | 0.215s | 44 | — |
| hf-011 | +71.05% | 0.907s | 0.262s | 47 | — |
| hf-012 | +74.52% | 0.784s | 0.200s | 43 | — |
| hf-013 | +71.08% | 0.762s | 0.221s | 41 | — |
| hf-014 | +62.90% | 0.777s | 0.288s | 43 | — |
| hf-016 | +75.29% | 0.882s | 0.218s | 49 | — |
| hf-017 | — | — | — | 17 | candidate output below tolerance |
| hf-015 | +72.97% | 1.194s | 0.323s | 72 | — |
| hf-018 | +73.40% | 0.782s | 0.208s | 47 | — |
| hf-019 | +71.36% | 0.697s | 0.200s | 42 | — |
| hf-020 | +72.14% | 0.661s | 0.184s | 42 | — |
| hf-021 | +73.55% | 0.723s | 0.191s | 45 | — |
| hf-022 | +64.19% | 0.608s | 0.218s | 41 | — |
| hf-023 | +73.53% | 0.747s | 0.198s | 44 | — |
| hf-024 | +71.66% | 0.743s | 0.211s | 43 | — |
| hf-025 | +76.23% | 0.684s | 0.163s | 42 | — |
| hf-026 | +66.01% | 0.601s | 0.204s | 41 | — |
| hf-028 | +74.73% | 0.517s | 0.131s | 29 | — |
| hf-027 | +73.07% | 0.705s | 0.190s | 43 | — |
| hf-029 | +77.19% | 0.790s | 0.180s | 51 | — |
| hf-030 | +72.55% | 0.664s | 0.182s | 47 | — |
| hf-031 | — | — | — | 29 | candidate output below tolerance |
| 43 |
| 49.96ms |
| — |
| hf-003 | 88.56ms | 0.197s | 40 | 21.19ms | — |
| hf-004 | 70.78ms | 0.198s | 47 | 18.28ms | — |
| hf-005 | 74.23ms | 0.298s | 69 | 46.54ms | — |
| hf-006 | 97.94ms | 0.192s | 51 | 16.65ms | — |
| hf-007 | 65.70ms | 0.209s | 67 | 18.14ms | — |
| hf-008 | 138.09ms | 0.307s | 46 | 46.93ms | — |
| hf-009 | 106.81ms | 0.264s | 43 | 46.80ms | — |
| hf-010 | 119.12ms | 0.215s | 44 | 17.73ms | — |
| hf-011 | 104.53ms | 0.262s | 47 | 46.85ms | — |
| hf-012 | 89.10ms | 0.200s | 43 | 18.44ms | — |
| hf-013 | 92.95ms | 0.221s | 42 | 18.00ms | — |
| hf-014 | 136.86ms | 0.288s | 43 | 44.65ms | — |
| hf-015 | 88.03ms | 0.323s | 72 | 47.40ms | — |
| hf-016 | 92.09ms | 0.266s | 53 | 48.50ms | — |
| hf-017 | 92.68ms | 0.141s | 17 | 17.86ms | — |
| hf-018 | 65.52ms | 0.208s | 48 | 18.11ms | — |
| hf-019 | 88.20ms | 0.200s | 42 | 18.14ms | — |
| hf-020 | 73.19ms | 0.184s | 42 | 17.96ms | — |
| hf-021 | 64.50ms | 0.191s | 49 | 17.30ms | — |
| hf-022 | 104.60ms | 0.218s | 41 | 18.64ms | — |
| hf-023 | 86.29ms | 0.213s | 48 | 17.27ms | — |
| hf-024 | 83.98ms | 0.211s | 43 | 21.39ms | — |
| hf-025 | 82.46ms | 0.163s | 42 | 17.63ms | — |
| hf-026 | 92.66ms | 0.204s | 41 | 17.75ms | — |
| hf-027 | 78.19ms | 0.190s | 43 | 18.60ms | — |
| hf-028 | 66.10ms | 0.131s | 29 | 17.60ms | — |
| hf-029 | 69.10ms | 0.180s | 51 | 18.10ms | — |
| hf-030 | 86.56ms | 0.182s | 47 | 18.03ms | — |
| hf-031 | 91.42ms | 0.234s | 29 | 17.30ms | — |
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