Round score
0.0799
+7.99%
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 | -3.50% | 0.736s | 0.761s | 48 | — |
| hf-001 | +5.48% | 0.723s | 0.683s | 42 | — |
| hf-002 | -0.78% | 0.722s | 0.728s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 102.43ms | 0.761s | 48 | 70.98ms | — |
| hf-001 | 81.43ms | 0.892s | 52 | 84.49ms | — |
| hf-002 | 74.21ms | 0.760s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 42 |
| — |
| hf-003 | +12.75% | 0.915s | 0.798s | 48 | — |
| hf-004 | — | — | — | 17 | candidate output below tolerance |
| hf-005 | +12.37% | 0.785s | 0.688s | 42 | — |
| hf-006 | — | — | — | 27 | candidate output below tolerance |
| hf-007 | +8.18% | 0.783s | 0.719s | 42 | — |
| hf-008 | +3.52% | 0.784s | 0.757s | 42 | — |
| hf-009 | +9.66% | 0.775s | 0.700s | 43 | — |
| hf-011 | +4.21% | 0.727s | 0.696s | 43 | — |
| hf-010 | +7.17% | 0.738s | 0.685s | 42 | — |
| hf-012 | +19.56% | 0.763s | 0.614s | 43 | — |
| hf-015 | +13.18% | 0.342s | 0.297s | 16 | — |
| hf-013 | — | — | — | 28 | candidate output below tolerance |
| hf-014 | +21.67% | 0.901s | 0.706s | 46 | — |
| hf-016 | +20.23% | 0.768s | 0.613s | 43 | — |
| hf-018 | +22.77% | 0.741s | 0.572s | 40 | — |
| hf-017 | +20.36% | 0.943s | 0.751s | 53 | — |
| hf-021 | +7.80% | 0.793s | 0.732s | 43 | — |
| hf-019 | +14.22% | 1.326s | 1.138s | 73 | — |
| hf-020 | — | — | — | 49 | candidate output below tolerance |
| hf-022 | +11.75% | 0.780s | 0.689s | 42 | — |
| hf-023 | +10.28% | 0.762s | 0.684s | 41 | — |
| hf-024 | — | — | — | 28 | candidate output below tolerance |
| hf-025 | +18.18% | 0.780s | 0.638s | 43 | — |
| hf-027 | +23.04% | 0.538s | 0.414s | 28 | — |
| hf-026 | +11.99% | 0.771s | 0.678s | 42 | — |
| hf-029 | +4.77% | 0.699s | 0.665s | 42 | — |
| hf-028 | +5.99% | 0.946s | 0.889s | 59 | — |
| hf-030 | +5.12% | 0.615s | 0.584s | 41 | — |
| hf-031 | +6.23% | 0.519s | 0.487s | 40 | — |
| 45 |
| 84.06ms |
| — |
| hf-003 | 97.09ms | 0.798s | 48 | 84.68ms | — |
| hf-004 | 120.83ms | 0.360s | 17 | 84.61ms | — |
| hf-005 | 95.41ms | 0.747s | 43 | 73.53ms | — |
| hf-006 | 80.30ms | 0.404s | 27 | 35.54ms | — |
| hf-007 | 81.00ms | 0.719s | 42 | 73.59ms | — |
| hf-008 | 146.86ms | 0.757s | 42 | 72.00ms | — |
| hf-009 | 73.28ms | 0.700s | 43 | 71.31ms | — |
| hf-010 | 75.60ms | 0.685s | 42 | 67.82ms | — |
| hf-011 | 112.40ms | 0.696s | 43 | 34.28ms | — |
| hf-012 | 75.70ms | 0.614s | 43 | 34.21ms | — |
| hf-013 | 74.59ms | 0.407s | 28 | 33.29ms | — |
| hf-014 | 74.37ms | 0.706s | 46 | 70.86ms | — |
| hf-015 | 93.13ms | 0.651s | 42 | 43.88ms | — |
| hf-016 | 75.69ms | 0.613s | 43 | 43.90ms | — |
| hf-017 | 73.52ms | 0.751s | 53 | 36.46ms | — |
| hf-018 | 79.88ms | 0.605s | 43 | 35.77ms | — |
| hf-019 | 63.04ms | 1.138s | 73 | 71.27ms | — |
| hf-020 | 69.50ms | 0.778s | 49 | 71.91ms | — |
| hf-021 | 106.15ms | 0.732s | 43 | 75.39ms | — |
| hf-022 | 84.89ms | 0.689s | 42 | 70.49ms | — |
| hf-023 | 99.28ms | 0.684s | 41 | 70.50ms | — |
| hf-024 | 92.53ms | 0.477s | 28 | 70.36ms | — |
| hf-025 | 67.36ms | 1.078s | 70 | 82.19ms | — |
| hf-026 | 86.49ms | 0.678s | 42 | 71.78ms | — |
| hf-027 | 77.68ms | 0.414s | 28 | 36.53ms | — |
| hf-028 | 74.55ms | 0.889s | 59 | 86.26ms | — |
| hf-029 | 99.59ms | 0.665s | 42 | 71.62ms | — |
| hf-030 | 111.94ms | 0.584s | 41 | 34.97ms | — |
| hf-031 | 74.53ms | 0.487s | 40 | 11.22ms | — |
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