Round score
0.7793
+77.93%
Prompts scored
25 / 32
median_e2e_speedup
Below tolerance
7
answered too little to score
Zeroed total
7
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 | +77.79% | 0.712s | 0.158s | 42 | — |
| hf-001 | +81.53% | 0.760s | 0.140s | 45 | — |
| hf-003 | +80.61% | 0.744s | 0.144s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 43.96ms | 0.158s | 42 | 17.54ms | — |
| hf-001 | 42.81ms | 0.140s | 45 | 16.53ms | — |
| hf-002 | 42.28ms | 0.139s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 43 |
| — |
| hf-002 | — | — | — | 51 | candidate output below tolerance |
| hf-004 | +81.06% | 0.730s | 0.138s | 42 | — |
| hf-005 | +79.31% | 0.753s | 0.156s | 47 | — |
| hf-006 | +72.19% | 0.694s | 0.193s | 42 | — |
| hf-007 | +72.31% | 0.671s | 0.186s | 43 | — |
| hf-009 | — | — | — | 21 | candidate output below tolerance |
| hf-008 | +80.28% | 0.942s | 0.186s | 63 | — |
| hf-010 | — | — | — | 29 | candidate output below tolerance |
| hf-012 | — | — | — | 24 | candidate output below tolerance |
| hf-011 | +80.92% | 0.903s | 0.172s | 52 | — |
| hf-013 | +78.93% | 0.497s | 0.105s | 28 | — |
| hf-014 | +82.42% | 0.739s | 0.130s | 42 | — |
| hf-015 | +78.30% | 0.717s | 0.156s | 41 | — |
| hf-016 | +80.32% | 0.739s | 0.145s | 43 | — |
| hf-017 | +78.23% | 0.742s | 0.162s | 43 | — |
| hf-018 | +82.41% | 0.704s | 0.124s | 42 | — |
| hf-019 | +82.08% | 0.872s | 0.156s | 54 | — |
| hf-020 | +77.98% | 0.662s | 0.146s | 41 | — |
| hf-021 | — | — | — | 21 | candidate output below tolerance |
| hf-023 | +77.87% | 0.764s | 0.169s | 52 | — |
| hf-022 | +80.27% | 0.655s | 0.129s | 42 | — |
| hf-024 | — | — | — | 41 | candidate output below tolerance |
| hf-026 | +70.20% | 0.359s | 0.107s | 23 | — |
| hf-025 | +79.44% | 0.591s | 0.122s | 42 | — |
| hf-029 | +72.66% | 0.325s | 0.089s | 20 | — |
| hf-028 | +77.72% | 0.708s | 0.158s | 41 | — |
| hf-027 | — | — | — | 36 | candidate output below tolerance |
| hf-030 | +77.67% | 0.586s | 0.131s | 41 | — |
| hf-031 | +61.27% | 0.297s | 0.115s | 20 | — |
| 51 |
| 16.37ms |
| — |
| hf-003 | 47.17ms | 0.144s | 43 | 16.93ms | — |
| hf-004 | 41.28ms | 0.138s | 42 | 16.50ms | — |
| hf-005 | 42.55ms | 0.156s | 48 | 16.46ms | — |
| hf-006 | 45.86ms | 0.369s | 100 | 53.45ms | — |
| hf-007 | 68.25ms | 0.186s | 43 | 17.26ms | — |
| hf-008 | 39.25ms | 0.186s | 66 | 16.90ms | — |
| hf-009 | 34.61ms | 0.147s | 21 | 16.71ms | — |
| hf-010 | 39.41ms | 0.121s | 29 | 16.42ms | — |
| hf-011 | 42.92ms | 0.188s | 69 | 16.52ms | — |
| hf-012 | 45.96ms | 0.111s | 24 | 16.21ms | — |
| hf-013 | 39.96ms | 0.105s | 28 | 16.70ms | — |
| hf-014 | 33.06ms | 0.130s | 48 | 16.68ms | — |
| hf-015 | 42.10ms | 0.172s | 52 | 16.82ms | — |
| hf-016 | 46.48ms | 0.145s | 43 | 17.04ms | — |
| hf-017 | 46.03ms | 0.162s | 43 | 17.76ms | — |
| hf-018 | 42.64ms | 0.124s | 43 | 16.58ms | — |
| hf-019 | 44.20ms | 0.156s | 54 | 16.58ms | — |
| hf-020 | 47.08ms | 0.146s | 41 | 17.75ms | — |
| hf-021 | 40.66ms | 0.073s | 21 | 16.29ms | — |
| hf-022 | 46.73ms | 0.129s | 42 | 17.12ms | — |
| hf-023 | 39.60ms | 0.169s | 53 | 16.36ms | — |
| hf-024 | 32.13ms | 0.145s | 41 | 16.68ms | — |
| hf-025 | 40.73ms | 0.122s | 42 | 16.48ms | — |
| hf-026 | 42.49ms | 0.156s | 43 | 16.56ms | — |
| hf-027 | 40.04ms | 0.136s | 36 | 16.73ms | — |
| hf-028 | 44.38ms | 0.158s | 41 | 16.43ms | — |
| hf-029 | 41.34ms | 0.138s | 46 | 16.71ms | — |
| hf-030 | 33.98ms | 0.131s | 42 | 16.43ms | — |
| hf-031 | 34.04ms | 0.115s | 20 | 16.76ms | — |
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