Round score
0.6222
+62.22%
Prompts scored
27 / 32
median_e2e_speedup
Below tolerance
5
answered too little to score
Zeroed total
5
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 | +67.34% | 0.766s | 0.250s | 49 | — |
| hf-002 | +60.93% | 0.723s | 0.283s | 40 | — |
| hf-005 | +60.40% | 0.395s | 0.156s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 73.03ms | 0.250s | 49 | 39.05ms | — |
| hf-001 | 90.97ms | 0.291s | 48 | 77.22ms | — |
| hf-002 | 123.31ms | 0.283s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 21 |
| — |
| hf-003 | +65.23% | 0.894s | 0.311s | 47 | — |
| hf-004 | +73.39% | 0.760s | 0.202s | 42 | — |
| hf-006 | +66.10% | 0.718s | 0.243s | 41 | — |
| hf-009 | +51.55% | 0.261s | 0.127s | 16 | — |
| hf-007 | — | — | — | 16 | candidate output below tolerance |
| hf-008 | +70.16% | 0.693s | 0.207s | 43 | — |
| hf-001 | — | — | — | 48 | candidate output below tolerance |
| hf-010 | +75.52% | 0.706s | 0.173s | 42 | — |
| hf-011 | +73.78% | 0.732s | 0.192s | 41 | — |
| hf-012 | +75.68% | 0.731s | 0.178s | 41 | — |
| hf-016 | +10.50% | 0.158s | 0.141s | 10 | — |
| hf-014 | +64.41% | 0.710s | 0.253s | 42 | — |
| hf-013 | +65.85% | 0.993s | 0.339s | 63 | — |
| hf-017 | +59.53% | 0.397s | 0.161s | 22 | — |
| hf-015 | +57.08% | 0.765s | 0.328s | 47 | — |
| hf-018 | +67.12% | 0.585s | 0.192s | 41 | — |
| hf-019 | +67.57% | 0.588s | 0.191s | 42 | — |
| hf-021 | +51.04% | 0.742s | 0.363s | 42 | — |
| hf-022 | +54.16% | 0.746s | 0.342s | 42 | — |
| hf-020 | +70.78% | 1.151s | 0.336s | 71 | — |
| hf-025 | +26.54% | 0.395s | 0.290s | 21 | — |
| hf-024 | +65.22% | 0.957s | 0.333s | 52 | — |
| hf-028 | +21.14% | 0.391s | 0.309s | 24 | — |
| hf-027 | +58.06% | 0.731s | 0.307s | 42 | — |
| hf-026 | — | — | — | 47 | candidate output below tolerance |
| hf-029 | +69.52% | 0.657s | 0.200s | 42 | — |
| hf-030 | — | — | — | 40 | candidate output below tolerance |
| hf-031 | +63.51% | 0.952s | 0.347s | 82 | — |
| hf-023 | — | — | — | 81 | candidate output below tolerance |
| 40 |
| 66.29ms |
| — |
| hf-003 | 104.81ms | 0.311s | 47 | 47.05ms | — |
| hf-004 | 94.12ms | 0.202s | 42 | 16.44ms | — |
| hf-005 | 78.73ms | 0.592s | 129 | 39.37ms | — |
| hf-006 | 95.25ms | 0.283s | 47 | 39.83ms | — |
| hf-007 | 83.04ms | 0.146s | 16 | 16.36ms | — |
| hf-008 | 68.48ms | 0.207s | 43 | 15.73ms | — |
| hf-009 | 78.82ms | 0.127s | 16 | 16.52ms | — |
| hf-010 | 65.56ms | 0.173s | 42 | 16.54ms | — |
| hf-011 | 68.09ms | 0.192s | 41 | 16.56ms | — |
| hf-012 | 69.62ms | 0.178s | 41 | 16.08ms | — |
| hf-013 | 55.53ms | 0.339s | 63 | 70.03ms | — |
| hf-014 | 106.88ms | 0.253s | 42 | 37.23ms | — |
| hf-015 | 95.26ms | 0.328s | 47 | 40.59ms | — |
| hf-016 | 94.78ms | 0.361s | 72 | 43.60ms | — |
| hf-017 | 97.33ms | 0.206s | 41 | 16.05ms | — |
| hf-018 | 84.05ms | 0.192s | 41 | 15.93ms | — |
| hf-019 | 67.54ms | 0.191s | 45 | 16.66ms | — |
| hf-020 | 67.45ms | 0.336s | 71 | 44.74ms | — |
| hf-021 | 89.96ms | 0.701s | 131 | 78.57ms | — |
| hf-022 | 130.25ms | 0.342s | 42 | 43.79ms | — |
| hf-023 | 89.16ms | 0.531s | 81 | 66.94ms | — |
| hf-024 | 84.03ms | 0.333s | 52 | 67.61ms | — |
| hf-025 | 114.62ms | 0.290s | 21 | 43.09ms | — |
| hf-026 | 90.30ms | 0.295s | 47 | 42.40ms | — |
| hf-027 | 95.01ms | 0.307s | 43 | 67.46ms | — |
| hf-028 | 106.70ms | 0.309s | 24 | 41.28ms | — |
| hf-029 | 92.18ms | 0.200s | 42 | 16.36ms | — |
| hf-030 | 103.72ms | 0.248s | 40 | 36.35ms | — |
| hf-031 | 88.30ms | 0.347s | 82 | 16.00ms | — |
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