Round score
-0.0001
-0.01%
Prompts scored
32 / 32
median_e2e_speedup
Below tolerance
0
answered too little to score
Zeroed total
0
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 | -0.16% | 0.489s | 0.489s | 28 | — |
| hf-001 | +0.01% | 0.871s | 0.871s | 52 | — |
| hf-002 | +0.02% | 0.788s | 0.788s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 96.81ms | 0.489s | 28 | 93.75ms | — |
| hf-001 | 105.48ms | 0.871s | 52 | 95.04ms | — |
| hf-002 | 67.82ms | 0.788s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 51 |
| — |
| hf-003 | +0.10% | 0.740s | 0.740s | 43 | — |
| hf-004 | +0.17% | 0.740s | 0.739s | 43 | — |
| hf-005 | +1.23% | 0.791s | 0.781s | 47 | — |
| hf-006 | +0.38% | 0.706s | 0.703s | 40 | — |
| hf-007 | -0.06% | 0.738s | 0.739s | 43 | — |
| hf-009 | -0.06% | 0.793s | 0.794s | 51 | — |
| hf-008 | +0.00% | 1.123s | 1.123s | 68 | — |
| hf-010 | -0.00% | 0.755s | 0.755s | 44 | — |
| hf-011 | -0.07% | 0.715s | 0.716s | 40 | — |
| hf-012 | -0.36% | 0.903s | 0.907s | 47 | — |
| hf-014 | -0.30% | 0.758s | 0.760s | 41 | — |
| hf-013 | -0.47% | 1.009s | 1.014s | 57 | — |
| hf-015 | -0.38% | 0.751s | 0.754s | 40 | — |
| hf-016 | +0.29% | 0.783s | 0.781s | 43 | — |
| hf-017 | -0.28% | 0.803s | 0.806s | 45 | — |
| hf-018 | -1.21% | 0.779s | 0.788s | 43 | — |
| hf-020 | -1.02% | 0.728s | 0.735s | 42 | — |
| hf-019 | +0.27% | 1.013s | 1.011s | 61 | — |
| hf-021 | -0.59% | 0.704s | 0.708s | 43 | — |
| hf-022 | +0.38% | 0.599s | 0.596s | 40 | — |
| hf-023 | +0.21% | 0.779s | 0.778s | 47 | — |
| hf-024 | +0.10% | 0.728s | 0.728s | 42 | — |
| hf-025 | -0.16% | 1.067s | 1.068s | 59 | — |
| hf-026 | -0.56% | 0.780s | 0.785s | 43 | — |
| hf-027 | -0.01% | 0.799s | 0.799s | 48 | — |
| hf-029 | +0.04% | 0.645s | 0.645s | 41 | — |
| hf-028 | -0.68% | 0.848s | 0.854s | 53 | — |
| hf-030 | +0.32% | 0.634s | 0.632s | 47 | — |
| hf-031 | +0.10% | 0.517s | 0.516s | 40 | — |
| 51 |
| 95.07ms |
| — |
| hf-003 | 103.51ms | 0.740s | 43 | 95.40ms | — |
| hf-004 | 105.91ms | 0.739s | 43 | 95.38ms | — |
| hf-005 | 70.89ms | 0.792s | 48 | 95.28ms | — |
| hf-006 | 106.58ms | 0.703s | 40 | 91.45ms | — |
| hf-007 | 107.20ms | 0.739s | 43 | 95.43ms | — |
| hf-008 | 106.82ms | 1.123s | 68 | 95.46ms | — |
| hf-009 | 71.04ms | 0.794s | 51 | 95.36ms | — |
| hf-010 | 106.88ms | 0.755s | 44 | 95.43ms | — |
| hf-011 | 70.67ms | 0.954s | 55 | 95.49ms | — |
| hf-012 | 106.47ms | 0.907s | 47 | 95.69ms | — |
| hf-013 | 112.31ms | 1.014s | 57 | 95.73ms | — |
| hf-014 | 105.85ms | 0.760s | 41 | 95.79ms | — |
| hf-015 | 111.72ms | 0.754s | 40 | 95.73ms | — |
| hf-016 | 106.81ms | 0.781s | 43 | 95.77ms | — |
| hf-017 | 106.19ms | 0.806s | 45 | 95.74ms | — |
| hf-018 | 111.98ms | 0.788s | 43 | 95.28ms | — |
| hf-019 | 104.71ms | 1.011s | 61 | 95.34ms | — |
| hf-020 | 109.89ms | 0.735s | 42 | 95.27ms | — |
| hf-021 | 105.78ms | 0.708s | 43 | 99.29ms | — |
| hf-022 | 67.30ms | 0.596s | 40 | 95.14ms | — |
| hf-023 | 103.56ms | 0.778s | 47 | 95.26ms | — |
| hf-024 | 65.62ms | 0.728s | 42 | 98.00ms | — |
| hf-025 | 100.15ms | 1.068s | 59 | 95.43ms | — |
| hf-026 | 112.00ms | 0.807s | 45 | 95.01ms | — |
| hf-027 | 106.83ms | 0.799s | 48 | 95.28ms | — |
| hf-028 | 109.76ms | 0.854s | 53 | 95.24ms | — |
| hf-029 | 104.13ms | 0.645s | 41 | 91.42ms | — |
| hf-030 | 69.52ms | 0.632s | 47 | 91.11ms | — |
| hf-031 | 104.30ms | 0.516s | 40 | 12.50ms | — |
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