Round score
0.1088
+10.88%
Prompts scored
28 / 32
median_e2e_speedup
Below tolerance
4
answered too little to score
Zeroed total
4
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 | +5.35% | 0.591s | 0.560s | 43 | — |
| hf-001 | +10.94% | 0.666s | 0.593s | 45 | — |
| hf-002 | +5.30% | 0.595s | 0.564s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 55.69ms | 0.560s | 43 | 51.52ms | — |
| hf-001 | 64.05ms | 0.593s | 45 | 51.74ms | — |
| hf-002 | 64.22ms | 0.564s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 42 |
| — |
| hf-003 | +0.64% | 0.576s | 0.573s | 43 | — |
| hf-004 | +0.27% | 0.547s | 0.546s | 40 | — |
| hf-005 | +4.45% | 0.594s | 0.568s | 42 | — |
| hf-006 | +18.24% | 0.923s | 0.755s | 56 | — |
| hf-007 | +22.88% | 0.734s | 0.566s | 41 | — |
| hf-008 | +20.80% | 0.918s | 0.727s | 52 | — |
| hf-009 | +22.47% | 0.749s | 0.581s | 42 | — |
| hf-010 | +24.51% | 0.731s | 0.551s | 40 | — |
| hf-011 | +19.40% | 0.864s | 0.697s | 50 | — |
| hf-012 | +18.82% | 0.704s | 0.571s | 42 | — |
| hf-013 | — | — | — | 41 | candidate output below tolerance |
| hf-015 | +17.14% | 0.713s | 0.591s | 43 | — |
| hf-014 | — | — | — | 47 | candidate output below tolerance |
| hf-016 | — | — | — | 43 | candidate output below tolerance |
| hf-017 | +14.80% | 0.805s | 0.686s | 48 | — |
| hf-018 | +13.68% | 0.771s | 0.666s | 48 | — |
| hf-019 | +14.18% | 0.666s | 0.571s | 42 | — |
| hf-020 | +4.89% | 0.600s | 0.571s | 42 | — |
| hf-021 | +5.33% | 0.572s | 0.542s | 40 | — |
| hf-022 | +19.16% | 0.696s | 0.563s | 42 | — |
| hf-024 | +18.51% | 0.708s | 0.577s | 42 | — |
| hf-023 | — | — | — | 41 | candidate output below tolerance |
| hf-025 | +18.03% | 0.898s | 0.737s | 53 | — |
| hf-026 | +11.54% | 0.782s | 0.691s | 48 | — |
| hf-027 | +9.25% | 0.729s | 0.662s | 47 | — |
| hf-029 | +10.83% | 0.655s | 0.584s | 43 | — |
| hf-028 | +9.87% | 0.941s | 0.848s | 64 | — |
| hf-030 | +6.31% | 0.615s | 0.576s | 46 | — |
| hf-031 | +5.77% | 0.539s | 0.508s | 43 | — |
| 42 |
| 51.58ms |
| — |
| hf-003 | 59.74ms | 0.573s | 43 | 52.64ms | — |
| hf-004 | 63.35ms | 0.546s | 40 | 54.72ms | — |
| hf-005 | 66.45ms | 0.568s | 42 | 54.49ms | — |
| hf-006 | 62.62ms | 0.755s | 56 | 52.51ms | — |
| hf-007 | 64.67ms | 0.566s | 41 | 53.49ms | — |
| hf-008 | 70.60ms | 0.727s | 52 | 55.50ms | — |
| hf-009 | 66.00ms | 0.581s | 42 | 55.46ms | — |
| hf-010 | 63.98ms | 0.562s | 41 | 55.35ms | — |
| hf-011 | 73.43ms | 0.697s | 50 | 54.45ms | — |
| hf-012 | 64.82ms | 0.709s | 52 | 54.36ms | — |
| hf-013 | 63.94ms | 0.558s | 41 | 54.07ms | — |
| hf-014 | 65.13ms | 0.662s | 47 | 53.71ms | — |
| hf-015 | 61.87ms | 0.591s | 43 | 51.79ms | — |
| hf-016 | 59.02ms | 0.591s | 43 | 53.87ms | — |
| hf-017 | 76.98ms | 0.686s | 48 | 53.84ms | — |
| hf-018 | 61.56ms | 0.666s | 48 | 51.81ms | — |
| hf-019 | 57.02ms | 0.571s | 42 | 56.31ms | — |
| hf-020 | 71.71ms | 0.571s | 42 | 49.74ms | — |
| hf-021 | 62.97ms | 0.542s | 40 | 51.83ms | — |
| hf-022 | 64.50ms | 0.563s | 42 | 51.98ms | — |
| hf-023 | 60.08ms | 0.557s | 41 | 58.00ms | — |
| hf-024 | 61.44ms | 0.577s | 42 | 50.06ms | — |
| hf-025 | 62.43ms | 0.737s | 53 | 49.49ms | — |
| hf-026 | 80.09ms | 0.691s | 48 | 51.23ms | — |
| hf-027 | 62.20ms | 0.662s | 47 | 50.99ms | — |
| hf-028 | 71.78ms | 0.848s | 64 | 51.09ms | — |
| hf-029 | 69.91ms | 0.584s | 43 | 50.77ms | — |
| hf-030 | 62.10ms | 0.576s | 46 | 50.99ms | — |
| hf-031 | 64.77ms | 0.508s | 43 | 11.09ms | — |
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