Round score
0.8092
+80.92%
Prompts scored
30 / 32
median_e2e_speedup
Below tolerance
2
answered too little to score
Zeroed total
2
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 | +70.57% | 0.384s | 0.113s | 21 | — |
| hf-001 | +82.91% | 0.715s | 0.122s | 41 | — |
| hf-002 | +82.08% | 0.761s | 0.136s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 54.66ms | 0.157s | 44 | 14.89ms | — |
| hf-001 | 47.80ms | 0.137s | 43 | 15.85ms | — |
| hf-002 | 35.10ms | 0.136s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 45 |
| — |
| hf-003 | +80.90% | 0.782s | 0.149s | 45 | — |
| hf-004 | +81.54% | 0.754s | 0.139s | 43 | — |
| hf-005 | +78.61% | 0.737s | 0.158s | 42 | — |
| hf-006 | +83.30% | 0.730s | 0.122s | 42 | — |
| hf-007 | +84.11% | 0.759s | 0.121s | 44 | — |
| hf-008 | +84.73% | 0.906s | 0.138s | 50 | — |
| hf-009 | +85.12% | 0.730s | 0.109s | 41 | — |
| hf-010 | +80.95% | 0.788s | 0.150s | 47 | — |
| hf-011 | — | — | — | 19 | candidate output below tolerance |
| hf-012 | +79.45% | 0.951s | 0.195s | 50 | — |
| hf-013 | +79.13% | 0.749s | 0.156s | 40 | — |
| hf-014 | +86.33% | 0.903s | 0.123s | 47 | — |
| hf-017 | +82.04% | 0.551s | 0.099s | 28 | — |
| hf-015 | +80.13% | 0.949s | 0.189s | 51 | — |
| hf-016 | +86.17% | 0.781s | 0.108s | 42 | — |
| hf-018 | +81.98% | 0.760s | 0.137s | 41 | — |
| hf-019 | +80.40% | 0.769s | 0.151s | 42 | — |
| hf-020 | +80.01% | 0.904s | 0.181s | 47 | — |
| hf-021 | +78.45% | 0.770s | 0.166s | 42 | — |
| hf-022 | +83.29% | 0.591s | 0.099s | 32 | — |
| hf-023 | +83.29% | 0.508s | 0.085s | 24 | — |
| hf-024 | +78.22% | 0.722s | 0.157s | 42 | — |
| hf-025 | +84.44% | 0.785s | 0.122s | 47 | — |
| hf-026 | +80.81% | 0.792s | 0.152s | 47 | — |
| hf-027 | +82.62% | 0.737s | 0.128s | 42 | — |
| hf-028 | +80.76% | 0.787s | 0.151s | 43 | — |
| hf-029 | +74.71% | 0.707s | 0.179s | 43 | — |
| hf-030 | +80.57% | 0.675s | 0.131s | 47 | — |
| hf-031 | — | — | — | 41 | candidate output below tolerance |
| 45 |
| 15.31ms |
| — |
| hf-003 | 46.55ms | 0.194s | 67 | 15.55ms | — |
| hf-004 | 37.65ms | 0.139s | 43 | 14.77ms | — |
| hf-005 | 41.44ms | 0.158s | 42 | 14.89ms | — |
| hf-006 | 34.80ms | 0.122s | 42 | 14.86ms | — |
| hf-007 | 47.27ms | 0.121s | 44 | 14.84ms | — |
| hf-008 | 36.13ms | 0.138s | 50 | 15.22ms | — |
| hf-009 | 36.82ms | 0.109s | 41 | 14.78ms | — |
| hf-010 | 46.15ms | 0.150s | 47 | 15.52ms | — |
| hf-011 | 47.59ms | 0.091s | 19 | 14.57ms | — |
| hf-012 | 47.95ms | 0.264s | 74 | 53.85ms | — |
| hf-013 | 69.26ms | 0.156s | 42 | 15.03ms | — |
| hf-014 | 36.29ms | 0.123s | 47 | 15.18ms | — |
| hf-015 | 42.33ms | 0.189s | 51 | 15.11ms | — |
| hf-016 | 36.07ms | 0.108s | 42 | 15.01ms | — |
| hf-017 | 40.76ms | 0.114s | 40 | 14.74ms | — |
| hf-018 | 35.95ms | 0.137s | 41 | 14.83ms | — |
| hf-019 | 35.28ms | 0.151s | 42 | 15.02ms | — |
| hf-020 | 47.25ms | 0.181s | 47 | 15.58ms | — |
| hf-021 | 47.39ms | 0.180s | 43 | 15.38ms | — |
| hf-022 | 40.44ms | 0.099s | 32 | 14.78ms | — |
| hf-023 | 41.08ms | 0.128s | 49 | 14.75ms | — |
| hf-024 | 40.99ms | 0.157s | 42 | 14.91ms | — |
| hf-025 | 35.37ms | 0.122s | 47 | 14.77ms | — |
| hf-026 | 36.08ms | 0.152s | 47 | 15.05ms | — |
| hf-027 | 40.63ms | 0.128s | 42 | 14.69ms | — |
| hf-028 | 35.49ms | 0.151s | 43 | 15.24ms | — |
| hf-029 | 33.80ms | 0.179s | 43 | 15.17ms | — |
| hf-030 | 43.72ms | 0.131s | 47 | 15.00ms | — |
| hf-031 | 34.58ms | 0.137s | 41 | 15.24ms | — |
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