Round score
0.6453
+64.53%
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 | +71.13% | 0.700s | 0.202s | 42 | — |
| hf-002 | +73.15% | 0.657s | 0.176s | 37 | — |
| hf-000 | — | — | — |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 106.35ms | 0.366s | 46 | 42.76ms | — |
| hf-001 | 94.90ms | 0.202s | 42 | 18.23ms | — |
| hf-002 | 72.35ms | 0.194s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 46 |
| candidate output below tolerance |
| hf-003 | +74.70% | 0.710s | 0.180s | 43 | — |
| hf-005 | +55.82% | 0.374s | 0.165s | 24 | — |
| hf-004 | +74.56% | 0.773s | 0.197s | 48 | — |
| hf-007 | +55.44% | 0.372s | 0.166s | 22 | — |
| hf-006 | +56.39% | 0.716s | 0.312s | 43 | — |
| hf-008 | +64.77% | 0.860s | 0.303s | 53 | — |
| hf-009 | +71.30% | 0.726s | 0.208s | 43 | — |
| hf-010 | +56.03% | 0.724s | 0.319s | 43 | — |
| hf-011 | +51.78% | 0.683s | 0.329s | 43 | — |
| hf-012 | +51.13% | 0.591s | 0.289s | 41 | — |
| hf-013 | +69.90% | 0.716s | 0.215s | 42 | — |
| hf-014 | +68.73% | 0.705s | 0.220s | 41 | — |
| hf-015 | +59.50% | 0.736s | 0.298s | 41 | — |
| hf-016 | +68.74% | 0.704s | 0.220s | 42 | — |
| hf-018 | +70.95% | 0.693s | 0.201s | 43 | — |
| hf-017 | +61.91% | 0.947s | 0.361s | 61 | — |
| hf-019 | +64.23% | 0.612s | 0.219s | 42 | — |
| hf-021 | +65.09% | 0.460s | 0.161s | 28 | — |
| hf-020 | — | — | — | 20 | candidate output below tolerance |
| hf-022 | +69.99% | 0.367s | 0.110s | 23 | — |
| hf-024 | +66.66% | 0.414s | 0.138s | 28 | — |
| hf-023 | — | — | — | 42 | candidate output below tolerance |
| hf-026 | +65.61% | 0.529s | 0.182s | 28 | — |
| hf-025 | +75.42% | 0.784s | 0.193s | 46 | — |
| hf-027 | +71.70% | 0.716s | 0.203s | 43 | — |
| hf-029 | +36.85% | 0.302s | 0.190s | 17 | — |
| hf-028 | +51.90% | 0.612s | 0.294s | 40 | — |
| hf-030 | +51.53% | 0.615s | 0.298s | 47 | — |
| hf-031 | +64.29% | 0.762s | 0.272s | 68 | — |
| 45 |
| 17.78ms |
| — |
| hf-003 | 74.87ms | 0.180s | 43 | 17.94ms | — |
| hf-004 | 57.84ms | 0.197s | 48 | 17.96ms | — |
| hf-005 | 76.84ms | 0.291s | 49 | 74.22ms | — |
| hf-006 | 129.26ms | 0.312s | 43 | 68.52ms | — |
| hf-007 | 112.22ms | 0.218s | 43 | 18.25ms | — |
| hf-008 | 94.55ms | 0.303s | 53 | 45.71ms | — |
| hf-009 | 103.04ms | 0.283s | 46 | 74.40ms | — |
| hf-010 | 113.38ms | 0.319s | 43 | 41.43ms | — |
| hf-011 | 99.98ms | 0.329s | 43 | 80.49ms | — |
| hf-012 | 143.89ms | 0.289s | 41 | 73.51ms | — |
| hf-013 | 113.54ms | 0.215s | 42 | 18.23ms | — |
| hf-014 | 130.78ms | 0.220s | 41 | 18.40ms | — |
| hf-015 | 107.59ms | 0.316s | 46 | 44.11ms | — |
| hf-016 | 98.18ms | 0.220s | 42 | 17.99ms | — |
| hf-017 | 113.88ms | 0.361s | 61 | 43.64ms | — |
| hf-018 | 94.43ms | 0.218s | 45 | 18.25ms | — |
| hf-019 | 96.44ms | 0.219s | 42 | 17.67ms | — |
| hf-020 | 118.01ms | 0.247s | 20 | 40.97ms | — |
| hf-021 | 89.56ms | 0.161s | 28 | 18.47ms | — |
| hf-022 | 58.10ms | 0.110s | 23 | 17.66ms | — |
| hf-023 | 67.33ms | 0.173s | 42 | 18.16ms | — |
| hf-024 | 67.76ms | 0.156s | 42 | 18.13ms | — |
| hf-025 | 69.99ms | 0.193s | 46 | 18.04ms | — |
| hf-026 | 76.64ms | 0.182s | 28 | 17.75ms | — |
| hf-027 | 80.00ms | 0.220s | 47 | 18.33ms | — |
| hf-028 | 108.40ms | 0.338s | 45 | 80.56ms | — |
| hf-029 | 137.44ms | 0.328s | 43 | 43.42ms | — |
| hf-030 | 91.02ms | 0.298s | 47 | 45.31ms | — |
| hf-031 | 97.97ms | 0.272s | 68 | 17.88ms | — |
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