Round score
0.7238
+72.38%
Prompts scored
29 / 32
median_e2e_speedup
Below tolerance
3
answered too little to score
Zeroed total
3
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 | +65.57% | 0.726s | 0.250s | 42 | — |
| hf-003 | +72.27% | 0.706s | 0.196s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 99.49ms | 0.195s | 42 | 17.10ms | — |
| hf-001 | 132.21ms | 0.310s | 40 | 57.64ms | — |
| hf-002 | 109.87ms | 0.285s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 40 |
| — |
| hf-004 | +73.65% | 0.740s | 0.195s | 43 | — |
| hf-005 | +77.54% | 0.744s | 0.167s | 40 | — |
| hf-006 | +81.21% | 0.924s | 0.174s | 46 | — |
| hf-007 | +76.33% | 0.610s | 0.144s | 33 | — |
| hf-008 | +65.77% | 0.895s | 0.306s | 46 | — |
| hf-009 | +66.68% | 0.773s | 0.257s | 43 | — |
| hf-001 | — | — | — | 40 | candidate output below tolerance |
| hf-010 | +73.34% | 0.797s | 0.213s | 44 | — |
| hf-011 | +71.68% | 0.907s | 0.257s | 47 | — |
| hf-012 | +74.76% | 0.784s | 0.198s | 43 | — |
| hf-013 | +71.62% | 0.762s | 0.216s | 41 | — |
| hf-014 | +64.34% | 0.777s | 0.277s | 43 | — |
| hf-016 | +71.84% | 0.882s | 0.248s | 49 | — |
| hf-017 | — | — | — | 17 | candidate output below tolerance |
| hf-015 | +74.11% | 1.194s | 0.309s | 72 | — |
| hf-018 | +73.71% | 0.782s | 0.206s | 47 | — |
| hf-019 | +71.69% | 0.697s | 0.197s | 42 | — |
| hf-020 | +72.50% | 0.661s | 0.182s | 42 | — |
| hf-021 | +73.78% | 0.723s | 0.190s | 45 | — |
| hf-022 | +64.68% | 0.608s | 0.215s | 41 | — |
| hf-023 | +73.80% | 0.747s | 0.196s | 44 | — |
| hf-024 | +71.64% | 0.743s | 0.211s | 43 | — |
| hf-025 | +75.96% | 0.684s | 0.164s | 42 | — |
| hf-026 | +66.32% | 0.601s | 0.203s | 41 | — |
| hf-028 | +75.12% | 0.517s | 0.129s | 29 | — |
| hf-027 | +73.52% | 0.705s | 0.187s | 43 | — |
| hf-029 | +77.47% | 0.790s | 0.178s | 51 | — |
| hf-030 | +73.00% | 0.664s | 0.179s | 47 | — |
| hf-031 | +65.98% | 0.530s | 0.180s | 41 | — |
| 43 |
| 45.37ms |
| — |
| hf-003 | 85.58ms | 0.196s | 40 | 16.92ms | — |
| hf-004 | 68.93ms | 0.195s | 47 | 17.70ms | — |
| hf-005 | 70.99ms | 0.295s | 69 | 51.45ms | — |
| hf-006 | 94.90ms | 0.189s | 51 | 16.45ms | — |
| hf-007 | 65.24ms | 0.208s | 67 | 17.14ms | — |
| hf-008 | 143.59ms | 0.306s | 46 | 45.59ms | — |
| hf-009 | 105.58ms | 0.257s | 43 | 42.79ms | — |
| hf-010 | 116.85ms | 0.213s | 44 | 17.69ms | — |
| hf-011 | 101.59ms | 0.257s | 47 | 43.55ms | — |
| hf-012 | 87.34ms | 0.198s | 43 | 17.04ms | — |
| hf-013 | 90.35ms | 0.216s | 42 | 20.24ms | — |
| hf-014 | 127.92ms | 0.277s | 43 | 42.02ms | — |
| hf-015 | 76.90ms | 0.309s | 72 | 40.38ms | — |
| hf-016 | 93.00ms | 0.282s | 53 | 44.07ms | — |
| hf-017 | 82.87ms | 0.130s | 17 | 16.29ms | — |
| hf-018 | 63.94ms | 0.206s | 48 | 16.70ms | — |
| hf-019 | 85.52ms | 0.197s | 42 | 18.08ms | — |
| hf-020 | 70.61ms | 0.182s | 42 | 17.46ms | — |
| hf-021 | 63.51ms | 0.190s | 49 | 18.46ms | — |
| hf-022 | 102.52ms | 0.215s | 41 | 17.89ms | — |
| hf-023 | 85.86ms | 0.211s | 48 | 17.03ms | — |
| hf-024 | 84.22ms | 0.211s | 44 | 18.55ms | — |
| hf-025 | 81.78ms | 0.164s | 42 | 22.84ms | — |
| hf-026 | 90.92ms | 0.203s | 41 | 17.75ms | — |
| hf-027 | 75.45ms | 0.187s | 43 | 17.44ms | — |
| hf-028 | 64.39ms | 0.129s | 29 | 17.38ms | — |
| hf-029 | 67.62ms | 0.178s | 51 | 21.16ms | — |
| hf-030 | 84.32ms | 0.179s | 47 | 17.16ms | — |
| hf-031 | 84.59ms | 0.180s | 42 | 17.28ms | — |
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