Round score
-0.0669
-6.69%
Prompts scored
23 / 32
median_e2e_speedup
Below tolerance
9
answered too little to score
Zeroed total
9
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 | -33.48% | 1.627s | 2.172s | 25 | — |
| hf-000 | -33.13% | 2.748s | 3.659s | 40 | — |
| hf-003 | -66.92% | 2.182s | 3.643s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 432.59ms | 4.260s | 43 | 235.68ms | — |
| hf-001 | 426.09ms | 2.172s | 25 | 234.10ms | — |
| hf-002 | 426.21ms | 6.271s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 42 |
| — |
| hf-005 | — | — | — | 28 | candidate output below tolerance |
| hf-004 | -82.64% | 1.991s | 3.636s | 42 | — |
| hf-002 | -23.87% | 3.417s | 4.233s | 52 | — |
| hf-006 | -48.55% | 2.468s | 3.666s | 41 | — |
| hf-007 | -20.05% | 3.198s | 3.839s | 42 | — |
| hf-008 | -34.65% | 3.020s | 4.066s | 43 | — |
| hf-011 | -31.22% | 2.958s | 3.882s | 42 | — |
| hf-009 | — | — | — | 37 | candidate output below tolerance |
| hf-012 | -1.68% | 4.107s | 4.176s | 43 | — |
| hf-013 | -28.95% | 3.019s | 3.893s | 41 | — |
| hf-014 | -7.15% | 3.653s | 3.914s | 48 | — |
| hf-023 | -7.56% | 2.086s | 2.243s | 42 | — |
| hf-017 | -6.67% | 3.101s | 3.308s | 43 | — |
| hf-020 | -6.49% | 2.541s | 2.706s | 41 | — |
| hf-015 | — | — | — | 52 | candidate output below tolerance |
| hf-022 | -7.22% | 2.233s | 2.395s | 46 | — |
| hf-019 | -5.50% | 2.872s | 3.030s | 52 | — |
| hf-024 | -4.70% | 1.920s | 2.010s | 43 | — |
| hf-025 | — | — | — | 30 | candidate output below tolerance |
| hf-021 | — | — | — | 47 | candidate output below tolerance |
| hf-027 | -7.12% | 1.434s | 1.536s | 42 | — |
| hf-010 | — | — | — | 42 | candidate output below tolerance |
| hf-018 | — | — | — | 53 | candidate output below tolerance |
| hf-029 | -6.85% | 1.062s | 1.135s | 40 | — |
| hf-030 | -6.71% | 0.906s | 0.967s | 42 | — |
| hf-026 | -6.08% | 1.737s | 1.842s | 53 | — |
| hf-031 | — | — | — | 46 | candidate output below tolerance |
| hf-016 | -4.60% | 3.817s | 3.993s | 89 | — |
| hf-028 | — | — | — | 49 | candidate output below tolerance |
| 81 |
| 236.28ms |
| — |
| hf-003 | 423.88ms | 3.835s | 43 | 239.60ms | — |
| hf-004 | 225.01ms | 3.636s | 42 | 244.20ms | — |
| hf-005 | 144.78ms | 1.784s | 28 | 235.14ms | — |
| hf-006 | 178.97ms | 3.666s | 41 | 238.26ms | — |
| hf-007 | 451.92ms | 3.839s | 42 | 233.55ms | — |
| hf-008 | 452.54ms | 4.066s | 43 | 234.19ms | — |
| hf-009 | 448.06ms | 3.399s | 37 | 246.83ms | — |
| hf-010 | 363.40ms | 3.853s | 42 | 241.55ms | — |
| hf-011 | 162.97ms | 4.279s | 44 | 236.40ms | — |
| hf-012 | 384.19ms | 4.176s | 43 | 237.72ms | — |
| hf-013 | 182.91ms | 3.893s | 41 | 246.05ms | — |
| hf-014 | 472.41ms | 3.914s | 48 | 229.99ms | — |
| hf-015 | 459.06ms | 3.779s | 52 | 229.90ms | — |
| hf-016 | 450.32ms | 3.993s | 89 | 231.51ms | — |
| hf-017 | 463.45ms | 3.308s | 43 | 236.91ms | — |
| hf-018 | 461.24ms | 3.242s | 53 | 233.82ms | — |
| hf-019 | 438.39ms | 3.130s | 61 | 229.36ms | — |
| hf-020 | 441.40ms | 2.706s | 41 | 233.30ms | — |
| hf-021 | 349.09ms | 2.591s | 47 | 232.28ms | — |
| hf-022 | 157.76ms | 2.395s | 46 | 230.11ms | — |
| hf-023 | 477.49ms | 2.280s | 45 | 235.16ms | — |
| hf-024 | 378.85ms | 2.222s | 62 | 222.04ms | — |
| hf-025 | 184.45ms | 1.720s | 30 | 224.79ms | — |
| hf-026 | 474.33ms | 1.842s | 53 | 212.33ms | — |
| hf-027 | 469.31ms | 1.536s | 42 | 208.74ms | — |
| hf-028 | 483.73ms | 1.430s | 49 | 208.42ms | — |
| hf-029 | 482.49ms | 1.135s | 40 | 187.68ms | — |
| hf-030 | 470.99ms | 0.967s | 42 | 20.76ms | — |
| hf-031 | 281.79ms | 0.829s | 46 | 18.57ms | — |
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