Round score
0.4229
+42.29%
Prompts scored
24 / 32
median_e2e_speedup
Below tolerance
8
answered too little to score
Zeroed total
8
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 | +49.78% | 2.407s | 1.209s | 41 | — |
| hf-002 | — | — | — | 28 | candidate output below tolerance |
| hf-008 | +30.53% | 2.028s | 1.409s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 196.77ms | 1.209s | 41 | 183.88ms | — |
| hf-001 | 227.35ms | 1.408s | 51 | 189.38ms | — |
| hf-002 | 234.84ms | 1.025s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 29 |
| — |
| hf-004 | +55.22% | 3.199s | 1.432s | 43 | — |
| hf-005 | +56.01% | 3.207s | 1.411s | 43 | — |
| hf-006 | — | — | — | 28 | candidate output below tolerance |
| hf-001 | +50.02% | 2.020s | 1.009s | 34 | — |
| hf-011 | — | — | — | 31 | candidate output below tolerance |
| hf-007 | +58.34% | 4.424s | 1.843s | 61 | — |
| hf-012 | — | — | — | 42 | candidate output below tolerance |
| hf-015 | +63.75% | 3.415s | 1.238s | 43 | — |
| hf-017 | +60.30% | 3.087s | 1.225s | 42 | — |
| hf-016 | +62.31% | 3.295s | 1.242s | 45 | — |
| hf-013 | +57.95% | 3.857s | 1.622s | 50 | — |
| hf-019 | +55.66% | 2.729s | 1.210s | 40 | — |
| hf-018 | +44.17% | 2.895s | 1.616s | 40 | — |
| hf-009 | +53.58% | 4.811s | 2.234s | 75 | — |
| hf-021 | +42.20% | 2.441s | 1.411s | 42 | — |
| hf-022 | +45.24% | 2.265s | 1.240s | 41 | — |
| hf-023 | +41.82% | 2.103s | 1.224s | 41 | — |
| hf-025 | +31.85% | 1.798s | 1.225s | 41 | — |
| hf-014 | — | — | — | 25 | candidate output below tolerance |
| hf-020 | +42.37% | 2.849s | 1.642s | 58 | — |
| hf-026 | +21.48% | 1.642s | 1.289s | 43 | — |
| hf-024 | — | — | — | 25 | candidate output below tolerance |
| hf-027 | +26.41% | 1.513s | 1.113s | 42 | — |
| hf-030 | +40.72% | 1.035s | 0.613s | 43 | — |
| hf-029 | +38.65% | 1.249s | 0.766s | 46 | — |
| hf-031 | +47.65% | 0.878s | 0.460s | 42 | — |
| hf-028 | — | — | — | 28 | candidate output below tolerance |
| hf-010 | +45.84% | 3.377s | 1.829s | 46 | — |
| hf-003 | — | — | — | 61 | candidate output below tolerance |
| 28 |
| 190.16ms |
| — |
| hf-003 | 242.46ms | 1.842s | 61 | 189.72ms | — |
| hf-004 | 408.12ms | 1.432s | 43 | 186.69ms | — |
| hf-005 | 233.06ms | 1.411s | 43 | 185.76ms | — |
| hf-006 | 422.51ms | 1.245s | 28 | 186.32ms | — |
| hf-007 | 405.19ms | 1.843s | 61 | 186.04ms | — |
| hf-008 | 231.74ms | 2.007s | 57 | 186.18ms | — |
| hf-009 | 246.10ms | 2.234s | 75 | 187.46ms | — |
| hf-010 | 418.11ms | 1.829s | 46 | 196.44ms | — |
| hf-011 | 241.75ms | 1.423s | 31 | 195.40ms | — |
| hf-012 | 247.80ms | 1.223s | 42 | 194.93ms | — |
| hf-013 | 413.66ms | 1.622s | 50 | 189.57ms | — |
| hf-014 | 234.14ms | 0.834s | 25 | 181.74ms | — |
| hf-015 | 244.31ms | 1.238s | 43 | 188.84ms | — |
| hf-016 | 241.01ms | 1.242s | 45 | 189.60ms | — |
| hf-017 | 232.19ms | 1.225s | 43 | 188.28ms | — |
| hf-018 | 409.93ms | 1.646s | 47 | 189.27ms | — |
| hf-019 | 237.91ms | 1.210s | 40 | 190.22ms | — |
| hf-020 | 248.57ms | 1.642s | 58 | 187.70ms | — |
| hf-021 | 244.43ms | 1.411s | 42 | 193.74ms | — |
| hf-022 | 249.49ms | 1.240s | 41 | 185.78ms | — |
| hf-023 | 416.01ms | 1.224s | 41 | 183.69ms | — |
| hf-024 | 241.82ms | 0.843s | 25 | 183.81ms | — |
| hf-025 | 403.70ms | 1.225s | 41 | 191.31ms | — |
| hf-026 | 413.53ms | 1.289s | 43 | 191.06ms | — |
| hf-027 | 238.92ms | 1.113s | 42 | 192.35ms | — |
| hf-028 | 408.00ms | 0.870s | 28 | 192.24ms | — |
| hf-029 | 411.26ms | 0.766s | 46 | 178.31ms | — |
| hf-030 | 237.48ms | 0.637s | 47 | 178.53ms | — |
| hf-031 | 246.35ms | 0.483s | 47 | 24.63ms | — |
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