Round score
-0.0033
-0.33%
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 | -0.36% | 0.489s | 0.490s | 28 | — |
| hf-001 | -0.22% | 0.871s | 0.873s | 52 | — |
| hf-002 | -0.41% | 0.788s | 0.791s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 96.57ms | 0.490s | 28 | 93.30ms | — |
| hf-001 | 105.94ms | 0.873s | 52 | 95.06ms | — |
| hf-002 | 68.75ms | 0.791s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 51 |
| — |
| hf-003 | -0.43% | 0.740s | 0.744s | 43 | — |
| hf-004 | -0.30% | 0.740s | 0.742s | 43 | — |
| hf-005 | +1.93% | 0.791s | 0.776s | 47 | — |
| hf-006 | -0.42% | 0.706s | 0.708s | 40 | — |
| hf-007 | +0.51% | 0.738s | 0.734s | 43 | — |
| hf-009 | +0.70% | 0.793s | 0.788s | 51 | — |
| hf-008 | +0.27% | 1.123s | 1.120s | 68 | — |
| hf-010 | -1.09% | 0.755s | 0.763s | 44 | — |
| hf-011 | +0.22% | 0.715s | 0.714s | 40 | — |
| hf-012 | -0.57% | 0.935s | 0.941s | 50 | — |
| hf-014 | -0.79% | 0.758s | 0.764s | 41 | — |
| hf-013 | -0.31% | 1.009s | 1.013s | 57 | — |
| hf-015 | +0.68% | 0.751s | 0.746s | 40 | — |
| hf-016 | -0.95% | 0.783s | 0.791s | 43 | — |
| hf-017 | -0.26% | 0.803s | 0.806s | 45 | — |
| hf-018 | -1.27% | 0.779s | 0.789s | 43 | — |
| hf-020 | -1.26% | 0.728s | 0.737s | 42 | — |
| hf-019 | +0.04% | 1.013s | 1.013s | 61 | — |
| hf-021 | +0.21% | 0.704s | 0.702s | 43 | — |
| hf-022 | -0.01% | 0.599s | 0.599s | 40 | — |
| hf-023 | -0.28% | 0.779s | 0.782s | 47 | — |
| hf-024 | -0.39% | 0.728s | 0.731s | 42 | — |
| hf-025 | +0.15% | 1.067s | 1.065s | 59 | — |
| hf-026 | -0.37% | 0.780s | 0.783s | 43 | — |
| hf-027 | -0.53% | 0.799s | 0.803s | 48 | — |
| hf-029 | -0.62% | 0.645s | 0.649s | 41 | — |
| hf-028 | -0.44% | 0.804s | 0.808s | 49 | — |
| hf-030 | +0.56% | 0.634s | 0.630s | 47 | — |
| hf-031 | -0.83% | 0.537s | 0.542s | 42 | — |
| 51 |
| 95.02ms |
| — |
| hf-003 | 105.47ms | 0.744s | 43 | 95.12ms | — |
| hf-004 | 107.65ms | 0.742s | 43 | 95.38ms | — |
| hf-005 | 62.36ms | 0.787s | 48 | 95.09ms | — |
| hf-006 | 109.93ms | 0.708s | 40 | 92.14ms | — |
| hf-007 | 100.55ms | 0.734s | 43 | 95.48ms | — |
| hf-008 | 101.25ms | 1.120s | 68 | 95.46ms | — |
| hf-009 | 64.38ms | 0.788s | 51 | 94.92ms | — |
| hf-010 | 113.43ms | 0.763s | 44 | 95.21ms | — |
| hf-011 | 66.79ms | 0.759s | 44 | 95.55ms | — |
| hf-012 | 111.31ms | 0.963s | 52 | 95.34ms | — |
| hf-013 | 109.61ms | 1.013s | 57 | 95.26ms | — |
| hf-014 | 110.29ms | 0.764s | 41 | 94.84ms | — |
| hf-015 | 104.36ms | 0.779s | 43 | 94.86ms | — |
| hf-016 | 114.44ms | 0.791s | 43 | 95.28ms | — |
| hf-017 | 104.03ms | 0.806s | 45 | 95.24ms | — |
| hf-018 | 110.10ms | 0.789s | 43 | 95.20ms | — |
| hf-019 | 104.37ms | 1.013s | 61 | 95.25ms | — |
| hf-020 | 109.18ms | 0.737s | 42 | 95.46ms | — |
| hf-021 | 103.78ms | 0.702s | 43 | 95.24ms | — |
| hf-022 | 68.48ms | 0.599s | 40 | 95.22ms | — |
| hf-023 | 105.33ms | 0.782s | 47 | 95.20ms | — |
| hf-024 | 70.68ms | 0.731s | 42 | 95.14ms | — |
| hf-025 | 105.32ms | 1.065s | 59 | 94.97ms | — |
| hf-026 | 106.67ms | 0.783s | 43 | 95.25ms | — |
| hf-027 | 109.45ms | 0.803s | 48 | 94.94ms | — |
| hf-028 | 104.75ms | 0.808s | 49 | 94.80ms | — |
| hf-029 | 107.39ms | 0.649s | 41 | 91.23ms | — |
| hf-030 | 65.32ms | 0.630s | 47 | 91.15ms | — |
| hf-031 | 106.59ms | 0.542s | 42 | 12.45ms | — |
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