Round score
0.6499
+64.99%
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-001 | +69.69% | 0.700s | 0.212s | 42 | — |
| hf-002 | +73.61% | 0.657s | 0.173s | 37 | — |
| hf-000 | — | — | — |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 103.11ms | 0.412s | 54 | 47.31ms | — |
| hf-001 | 95.91ms | 0.212s | 42 | 17.08ms | — |
| hf-002 | 77.92ms | 0.189s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 54 |
| candidate output below tolerance |
| hf-003 | +74.57% | 0.710s | 0.181s | 43 | — |
| hf-005 | +64.78% | 0.374s | 0.132s | 24 | — |
| hf-004 | +76.05% | 0.773s | 0.185s | 48 | — |
| hf-007 | +54.56% | 0.372s | 0.169s | 22 | — |
| hf-006 | +59.01% | 0.716s | 0.294s | 43 | — |
| hf-008 | +62.84% | 0.860s | 0.320s | 53 | — |
| hf-009 | +72.04% | 0.726s | 0.203s | 43 | — |
| hf-010 | +59.48% | 0.724s | 0.294s | 43 | — |
| hf-011 | +67.92% | 0.683s | 0.219s | 43 | — |
| hf-012 | +47.97% | 0.591s | 0.307s | 41 | — |
| hf-013 | +69.82% | 0.716s | 0.216s | 42 | — |
| hf-014 | +56.90% | 0.705s | 0.304s | 41 | — |
| hf-015 | +65.06% | 0.736s | 0.257s | 41 | — |
| hf-016 | +70.77% | 0.704s | 0.206s | 42 | — |
| hf-018 | +70.01% | 0.693s | 0.208s | 43 | — |
| hf-017 | +63.43% | 0.947s | 0.346s | 61 | — |
| hf-019 | +68.03% | 0.612s | 0.196s | 42 | — |
| hf-021 | +67.58% | 0.460s | 0.149s | 28 | — |
| hf-020 | — | — | — | 20 | candidate output below tolerance |
| hf-022 | +67.01% | 0.367s | 0.121s | 23 | — |
| hf-024 | +64.80% | 0.414s | 0.146s | 28 | — |
| hf-023 | — | — | — | 42 | candidate output below tolerance |
| hf-026 | +66.96% | 0.529s | 0.175s | 28 | — |
| hf-025 | +74.84% | 0.784s | 0.197s | 46 | — |
| hf-027 | +70.60% | 0.716s | 0.210s | 43 | — |
| hf-029 | +29.33% | 0.302s | 0.213s | 17 | — |
| hf-028 | +64.92% | 0.612s | 0.215s | 40 | — |
| hf-030 | +56.99% | 0.615s | 0.265s | 47 | — |
| hf-031 | +60.42% | 0.762s | 0.302s | 68 | — |
| 45 |
| 16.70ms |
| — |
| hf-003 | 67.75ms | 0.181s | 43 | 16.89ms | — |
| hf-004 | 56.66ms | 0.185s | 48 | 16.41ms | — |
| hf-005 | 66.69ms | 0.211s | 49 | 16.92ms | — |
| hf-006 | 114.10ms | 0.294s | 43 | 81.20ms | — |
| hf-007 | 103.22ms | 0.217s | 43 | 16.91ms | — |
| hf-008 | 91.29ms | 0.320s | 53 | 42.50ms | — |
| hf-009 | 89.80ms | 0.203s | 46 | 16.83ms | — |
| hf-010 | 102.50ms | 0.294s | 43 | 39.61ms | — |
| hf-011 | 91.21ms | 0.219s | 43 | 16.46ms | — |
| hf-012 | 128.15ms | 0.307s | 41 | 89.44ms | — |
| hf-013 | 107.22ms | 0.216s | 42 | 16.18ms | — |
| hf-014 | 128.74ms | 0.304s | 41 | 68.42ms | — |
| hf-015 | 103.62ms | 0.298s | 46 | 40.70ms | — |
| hf-016 | 92.66ms | 0.206s | 42 | 16.77ms | — |
| hf-017 | 100.74ms | 0.346s | 61 | 42.31ms | — |
| hf-018 | 93.64ms | 0.208s | 45 | 16.66ms | — |
| hf-019 | 82.98ms | 0.196s | 42 | 16.65ms | — |
| hf-020 | 93.48ms | 0.191s | 20 | 16.70ms | — |
| hf-021 | 68.53ms | 0.149s | 28 | 16.43ms | — |
| hf-022 | 56.94ms | 0.121s | 23 | 16.39ms | — |
| hf-023 | 66.37ms | 0.179s | 42 | 16.42ms | — |
| hf-024 | 65.19ms | 0.178s | 42 | 16.67ms | — |
| hf-025 | 68.75ms | 0.197s | 46 | 16.66ms | — |
| hf-026 | 78.55ms | 0.175s | 28 | 16.25ms | — |
| hf-027 | 80.39ms | 0.279s | 47 | 68.14ms | — |
| hf-028 | 101.84ms | 0.292s | 45 | 77.23ms | — |
| hf-029 | 130.97ms | 0.213s | 17 | 16.75ms | — |
| hf-030 | 86.47ms | 0.265s | 47 | 49.25ms | — |
| hf-031 | 93.97ms | 0.302s | 68 | 16.98ms | — |
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