Round score
0.5653
+56.53%
Prompts scored
29 / 32
median_e2e_speedup
Below tolerance
3
answered too little to score
Zeroed total
3
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 | +56.30% | 0.735s | 0.321s | 46 | — |
| hf-002 | +57.49% | 0.745s | 0.317s | 42 | — |
| hf-005 | +50.10% | 0.395s | 0.197s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 82.65ms | 0.321s | 46 | 74.53ms | — |
| hf-001 | 104.16ms | 0.810s | 134 | 100.38ms | — |
| hf-002 | 124.77ms | 0.317s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 21 |
| — |
| hf-003 | +61.87% | 0.894s | 0.341s | 47 | — |
| hf-004 | +63.21% | 0.760s | 0.280s | 42 | — |
| hf-006 | +59.76% | 0.718s | 0.289s | 41 | — |
| hf-009 | +42.60% | 0.261s | 0.150s | 16 | — |
| hf-007 | +56.51% | 0.689s | 0.300s | 42 | — |
| hf-008 | +56.52% | 0.693s | 0.302s | 43 | — |
| hf-001 | +62.70% | 2.171s | 0.810s | 134 | — |
| hf-010 | +68.93% | 0.706s | 0.219s | 42 | — |
| hf-011 | +60.16% | 0.732s | 0.292s | 41 | — |
| hf-012 | +61.27% | 0.731s | 0.283s | 41 | — |
| hf-016 | +17.03% | 0.158s | 0.131s | 10 | — |
| hf-014 | +53.74% | 0.710s | 0.328s | 42 | — |
| hf-013 | +59.07% | 0.993s | 0.406s | 63 | — |
| hf-017 | +61.51% | 0.397s | 0.153s | 22 | — |
| hf-015 | +55.76% | 0.786s | 0.348s | 49 | — |
| hf-018 | +63.46% | 0.585s | 0.214s | 41 | — |
| hf-019 | +52.25% | 0.588s | 0.281s | 42 | — |
| hf-021 | +45.50% | 0.742s | 0.404s | 42 | — |
| hf-022 | +53.34% | 0.746s | 0.348s | 42 | — |
| hf-020 | +52.54% | 1.107s | 0.525s | 67 | — |
| hf-025 | +60.77% | 0.492s | 0.193s | 23 | — |
| hf-024 | +62.41% | 0.957s | 0.360s | 52 | — |
| hf-028 | +49.04% | 0.391s | 0.199s | 24 | — |
| hf-027 | +56.53% | 0.731s | 0.318s | 42 | — |
| hf-026 | — | — | — | 23 | candidate output below tolerance |
| hf-029 | +68.05% | 0.657s | 0.210s | 42 | — |
| hf-030 | — | — | — | 40 | candidate output below tolerance |
| hf-031 | +58.51% | 0.952s | 0.395s | 82 | — |
| hf-023 | — | — | — | 47 | candidate output below tolerance |
| 42 |
| 98.12ms |
| — |
| hf-003 | 117.02ms | 0.341s | 47 | 70.57ms | — |
| hf-004 | 104.86ms | 0.280s | 42 | 70.13ms | — |
| hf-005 | 101.31ms | 0.316s | 47 | 65.08ms | — |
| hf-006 | 99.02ms | 0.297s | 47 | 71.96ms | — |
| hf-007 | 95.37ms | 0.300s | 42 | 66.19ms | — |
| hf-008 | 99.61ms | 0.302s | 43 | 65.79ms | — |
| hf-009 | 99.42ms | 0.285s | 43 | 67.94ms | — |
| hf-010 | 101.93ms | 0.219s | 42 | 16.97ms | — |
| hf-011 | 103.10ms | 0.292s | 41 | 70.61ms | — |
| hf-012 | 91.98ms | 0.283s | 41 | 74.51ms | — |
| hf-013 | 93.31ms | 0.406s | 63 | 106.66ms | — |
| hf-014 | 124.80ms | 0.328s | 42 | 75.55ms | — |
| hf-015 | 94.67ms | 0.365s | 55 | 74.26ms | — |
| hf-016 | 96.94ms | 0.199s | 27 | 17.06ms | — |
| hf-017 | 85.85ms | 0.203s | 41 | 17.31ms | — |
| hf-018 | 100.77ms | 0.214s | 41 | 17.04ms | — |
| hf-019 | 92.38ms | 0.281s | 42 | 68.70ms | — |
| hf-020 | 95.33ms | 0.525s | 67 | 104.47ms | — |
| hf-021 | 101.53ms | 1.117s | 192 | 104.69ms | — |
| hf-022 | 153.09ms | 0.348s | 42 | 68.27ms | — |
| hf-023 | 112.44ms | 0.372s | 47 | 65.42ms | — |
| hf-024 | 103.06ms | 0.360s | 52 | 111.06ms | — |
| hf-025 | 122.72ms | 0.317s | 44 | 66.29ms | — |
| hf-026 | 101.06ms | 0.204s | 23 | 17.76ms | — |
| hf-027 | 101.82ms | 0.318s | 42 | 92.17ms | — |
| hf-028 | 117.97ms | 0.216s | 28 | 16.77ms | — |
| hf-029 | 92.33ms | 0.210s | 42 | 17.07ms | — |
| hf-030 | 120.21ms | 0.307s | 40 | 69.79ms | — |
| hf-031 | 93.34ms | 0.395s | 83 | 21.77ms | — |
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