Round score
0.6119
+61.19%
Prompts scored
27 / 32
median_e2e_speedup
Below tolerance
5
answered too little to score
Zeroed total
5
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 | +63.82% | 0.766s | 0.277s | 49 | — |
| hf-002 | +58.95% | 0.745s | 0.306s | 42 | — |
| hf-005 | +59.28% | 0.395s | 0.161s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 78.24ms | 0.277s | 49 | 61.11ms | — |
| hf-001 | 100.97ms | 0.343s | 46 | 100.93ms | — |
| hf-002 | 125.23ms | 0.306s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 21 |
| — |
| hf-003 | +63.87% | 0.894s | 0.323s | 47 | — |
| hf-004 | +73.38% | 0.760s | 0.202s | 42 | — |
| hf-006 | +70.52% | 0.718s | 0.212s | 41 | — |
| hf-009 | +49.45% | 0.261s | 0.132s | 16 | — |
| hf-007 | — | — | — | 16 | candidate output below tolerance |
| hf-008 | +69.95% | 0.693s | 0.208s | 43 | — |
| hf-001 | — | — | — | 46 | candidate output below tolerance |
| hf-010 | +74.43% | 0.706s | 0.180s | 42 | — |
| hf-011 | +72.83% | 0.721s | 0.196s | 40 | — |
| hf-012 | +72.38% | 0.731s | 0.202s | 41 | — |
| hf-016 | +12.97% | 0.158s | 0.137s | 10 | — |
| hf-014 | +59.89% | 0.700s | 0.281s | 41 | — |
| hf-013 | — | — | — | 51 | candidate output below tolerance |
| hf-017 | +55.42% | 0.397s | 0.177s | 22 | — |
| hf-015 | +60.95% | 0.786s | 0.307s | 49 | — |
| hf-018 | +64.64% | 0.585s | 0.207s | 41 | — |
| hf-019 | +64.89% | 0.588s | 0.207s | 42 | — |
| hf-021 | +57.47% | 0.742s | 0.316s | 42 | — |
| hf-022 | +54.21% | 0.746s | 0.341s | 42 | — |
| hf-020 | — | — | — | 62 | candidate output below tolerance |
| hf-025 | +62.95% | 0.492s | 0.182s | 23 | — |
| hf-024 | +64.85% | 0.957s | 0.336s | 52 | — |
| hf-028 | +59.03% | 0.391s | 0.160s | 24 | — |
| hf-027 | +70.40% | 0.731s | 0.216s | 42 | — |
| hf-026 | +66.37% | 0.962s | 0.324s | 55 | — |
| hf-029 | +70.98% | 0.657s | 0.191s | 42 | — |
| hf-030 | +51.55% | 0.637s | 0.309s | 45 | — |
| hf-031 | +61.42% | 0.952s | 0.367s | 82 | — |
| hf-023 | — | — | — | 37 | candidate output below tolerance |
| 42 |
| 87.96ms |
| — |
| hf-003 | 112.17ms | 0.323s | 47 | 62.55ms | — |
| hf-004 | 94.87ms | 0.202s | 42 | 16.16ms | — |
| hf-005 | 83.37ms | 0.270s | 42 | 62.78ms | — |
| hf-006 | 89.26ms | 0.274s | 47 | 62.46ms | — |
| hf-007 | 83.78ms | 0.146s | 16 | 16.10ms | — |
| hf-008 | 71.09ms | 0.208s | 43 | 15.53ms | — |
| hf-009 | 85.70ms | 0.193s | 40 | 16.10ms | — |
| hf-010 | 73.20ms | 0.180s | 42 | 15.72ms | — |
| hf-011 | 73.34ms | 0.196s | 40 | 16.34ms | — |
| hf-012 | 79.51ms | 0.202s | 41 | 15.84ms | — |
| hf-013 | 72.77ms | 0.210s | 51 | 15.74ms | — |
| hf-014 | 109.08ms | 0.281s | 41 | 64.31ms | — |
| hf-015 | 102.17ms | 0.307s | 50 | 61.44ms | — |
| hf-016 | 90.95ms | 0.377s | 68 | 60.97ms | — |
| hf-017 | 98.81ms | 0.297s | 41 | 89.99ms | — |
| hf-018 | 103.31ms | 0.207s | 41 | 15.58ms | — |
| hf-019 | 83.39ms | 0.207s | 42 | 15.86ms | — |
| hf-020 | 82.50ms | 0.386s | 62 | 62.77ms | — |
| hf-021 | 92.24ms | 0.316s | 42 | 98.89ms | — |
| hf-022 | 145.22ms | 0.341s | 42 | 61.50ms | — |
| hf-023 | 93.32ms | 0.315s | 37 | 61.94ms | — |
| hf-024 | 98.49ms | 0.336s | 52 | 110.70ms | — |
| hf-025 | 123.11ms | 0.298s | 44 | 63.11ms | — |
| hf-026 | 98.03ms | 0.355s | 67 | 60.91ms | — |
| hf-027 | 92.29ms | 0.216s | 42 | 16.34ms | — |
| hf-028 | 98.40ms | 0.206s | 43 | 15.98ms | — |
| hf-029 | 82.91ms | 0.191s | 42 | 15.94ms | — |
| hf-030 | 105.36ms | 0.309s | 45 | 60.24ms | — |
| hf-031 | 92.44ms | 0.367s | 83 | 16.17ms | — |
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