Round score
0.6973
+69.73%
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 | +60.77% | 0.801s | 0.314s | 47 | — |
| hf-002 | +59.99% | 0.726s | 0.291s | 42 | — |
| hf-003 | +71.21% | 0.706s | 0.203s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 112.24ms | 0.314s | 47 | 103.20ms | — |
| hf-001 | 155.23ms | 0.354s | 47 | 72.04ms | — |
| hf-002 | 119.04ms | 0.307s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 40 |
| — |
| hf-004 | +75.50% | 0.740s | 0.181s | 43 | — |
| hf-005 | +72.31% | 0.744s | 0.206s | 40 | — |
| hf-006 | +78.56% | 0.924s | 0.198s | 46 | — |
| hf-007 | +65.64% | 0.610s | 0.210s | 33 | — |
| hf-008 | +63.85% | 0.895s | 0.323s | 46 | — |
| hf-009 | — | — | — | 29 | candidate output below tolerance |
| hf-001 | — | — | — | 47 | candidate output below tolerance |
| hf-010 | +74.93% | 0.797s | 0.200s | 44 | — |
| hf-011 | +68.62% | 0.907s | 0.285s | 47 | — |
| hf-012 | +72.84% | 0.784s | 0.213s | 43 | — |
| hf-013 | +58.20% | 0.752s | 0.314s | 40 | — |
| hf-014 | +57.61% | 0.777s | 0.329s | 43 | — |
| hf-016 | — | — | — | 23 | candidate output below tolerance |
| hf-017 | +75.28% | 0.762s | 0.188s | 45 | — |
| hf-015 | +74.01% | 1.194s | 0.310s | 72 | — |
| hf-018 | +76.38% | 0.782s | 0.185s | 47 | — |
| hf-019 | +71.70% | 0.697s | 0.197s | 42 | — |
| hf-020 | +70.84% | 0.661s | 0.193s | 42 | — |
| hf-021 | +74.02% | 0.723s | 0.188s | 45 | — |
| hf-022 | +51.37% | 0.608s | 0.296s | 41 | — |
| hf-023 | +76.01% | 0.747s | 0.179s | 44 | — |
| hf-024 | +57.69% | 0.743s | 0.315s | 43 | — |
| hf-025 | +73.67% | 0.684s | 0.180s | 42 | — |
| hf-026 | +64.33% | 0.601s | 0.215s | 41 | — |
| hf-028 | +73.36% | 0.517s | 0.138s | 29 | — |
| hf-027 | — | — | — | 26 | candidate output below tolerance |
| hf-029 | +76.74% | 0.811s | 0.189s | 53 | — |
| hf-030 | +67.72% | 0.664s | 0.214s | 47 | — |
| hf-031 | +59.12% | 0.530s | 0.217s | 41 | — |
| 48 |
| 71.45ms |
| — |
| hf-003 | 106.16ms | 0.203s | 40 | 16.77ms | — |
| hf-004 | 81.97ms | 0.181s | 43 | 18.67ms | — |
| hf-005 | 74.36ms | 0.334s | 66 | 64.76ms | — |
| hf-006 | 98.87ms | 0.215s | 51 | 17.56ms | — |
| hf-007 | 94.03ms | 0.360s | 43 | 103.90ms | — |
| hf-008 | 159.90ms | 0.323s | 51 | 65.73ms | — |
| hf-009 | 105.97ms | 0.189s | 29 | 17.53ms | — |
| hf-010 | 100.56ms | 0.200s | 44 | 18.20ms | — |
| hf-011 | 101.44ms | 0.285s | 47 | 68.28ms | — |
| hf-012 | 98.40ms | 0.213s | 43 | 17.86ms | — |
| hf-013 | 116.42ms | 0.314s | 40 | 98.27ms | — |
| hf-014 | 145.42ms | 0.329s | 43 | 63.40ms | — |
| hf-015 | 94.70ms | 0.310s | 72 | 74.05ms | — |
| hf-016 | 92.57ms | 0.143s | 23 | 19.79ms | — |
| hf-017 | 73.12ms | 0.188s | 45 | 18.15ms | — |
| hf-018 | 70.11ms | 0.185s | 47 | 17.92ms | — |
| hf-019 | 81.93ms | 0.197s | 42 | 18.36ms | — |
| hf-020 | 77.63ms | 0.193s | 42 | 18.22ms | — |
| hf-021 | 73.05ms | 0.204s | 50 | 18.12ms | — |
| hf-022 | 126.65ms | 0.296s | 41 | 73.51ms | — |
| hf-023 | 102.14ms | 0.291s | 64 | 75.30ms | — |
| hf-024 | 101.20ms | 0.315s | 43 | 72.03ms | — |
| hf-025 | 98.14ms | 0.180s | 42 | 16.44ms | — |
| hf-026 | 99.58ms | 0.215s | 41 | 17.75ms | — |
| hf-027 | 91.89ms | 0.175s | 26 | 17.37ms | — |
| hf-028 | 71.00ms | 0.138s | 29 | 17.86ms | — |
| hf-029 | 73.80ms | 0.189s | 53 | 18.53ms | — |
| hf-030 | 98.69ms | 0.214s | 51 | 18.59ms | — |
| hf-031 | 117.99ms | 0.217s | 42 | 17.84ms | — |
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