Round score
0.8104
+81.04%
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-001 | +76.30% | 0.543s | 0.129s | 31 | — |
| hf-000 | — | — | — | 38 | candidate output below tolerance |
| hf-002 | — | — | — |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 54.01ms | 0.172s | 38 | 15.58ms | — |
| hf-001 | 40.66ms | 0.129s | 31 | 14.98ms | — |
| hf-002 | 43.94ms | 0.088s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 24 |
| candidate output below tolerance |
| hf-004 | +79.70% | 0.767s | 0.156s | 41 | — |
| hf-003 | +81.76% | 1.151s | 0.210s | 66 | — |
| hf-005 | — | — | — | 43 | candidate output below tolerance |
| hf-007 | +84.11% | 0.970s | 0.154s | 49 | — |
| hf-006 | — | — | — | 18 | candidate output below tolerance |
| hf-008 | — | — | — | 27 | candidate output below tolerance |
| hf-009 | +82.65% | 0.958s | 0.166s | 48 | — |
| hf-010 | +84.52% | 0.925s | 0.143s | 46 | — |
| hf-011 | +82.68% | 0.795s | 0.138s | 40 | — |
| hf-013 | +84.75% | 0.807s | 0.123s | 42 | — |
| hf-015 | +79.08% | 0.516s | 0.108s | 22 | — |
| hf-012 | — | — | — | 43 | candidate output below tolerance |
| hf-014 | +83.01% | 0.812s | 0.138s | 42 | — |
| hf-016 | +84.97% | 0.526s | 0.079s | 24 | — |
| hf-017 | +83.20% | 0.790s | 0.133s | 41 | — |
| hf-018 | +85.64% | 0.900s | 0.129s | 48 | — |
| hf-021 | +78.69% | 0.329s | 0.070s | 17 | — |
| hf-019 | +81.60% | 0.724s | 0.133s | 41 | — |
| hf-020 | +83.03% | 0.715s | 0.121s | 41 | — |
| hf-022 | +77.59% | 0.756s | 0.169s | 48 | — |
| hf-023 | +74.28% | 0.704s | 0.181s | 44 | — |
| hf-024 | +79.71% | 0.706s | 0.143s | 40 | — |
| hf-025 | +82.39% | 0.772s | 0.136s | 43 | — |
| hf-026 | +82.12% | 0.759s | 0.136s | 42 | — |
| hf-027 | +82.94% | 0.762s | 0.130s | 41 | — |
| hf-028 | +83.86% | 0.798s | 0.129s | 44 | — |
| hf-029 | +77.97% | 0.697s | 0.153s | 42 | — |
| hf-030 | +80.49% | 0.705s | 0.138s | 50 | — |
| hf-031 | — | — | — | 23 | candidate output below tolerance |
| 24 |
| 14.82ms |
| — |
| hf-003 | 47.54ms | 0.210s | 66 | 15.34ms | — |
| hf-004 | 53.95ms | 0.156s | 41 | 15.01ms | — |
| hf-005 | 35.93ms | 0.123s | 43 | 14.80ms | — |
| hf-006 | 44.35ms | 0.074s | 18 | 14.87ms | — |
| hf-007 | 37.56ms | 0.154s | 49 | 15.21ms | — |
| hf-008 | 41.84ms | 0.144s | 27 | 15.08ms | — |
| hf-009 | 47.36ms | 0.166s | 51 | 15.69ms | — |
| hf-010 | 40.52ms | 0.143s | 46 | 15.08ms | — |
| hf-011 | 49.08ms | 0.138s | 40 | 15.31ms | — |
| hf-012 | 35.07ms | 0.123s | 43 | 15.34ms | — |
| hf-013 | 47.70ms | 0.123s | 42 | 15.33ms | — |
| hf-014 | 36.25ms | 0.138s | 42 | 14.92ms | — |
| hf-015 | 34.77ms | 0.151s | 43 | 15.15ms | — |
| hf-016 | 35.24ms | 0.094s | 29 | 14.93ms | — |
| hf-017 | 43.93ms | 0.133s | 41 | 15.71ms | — |
| hf-018 | 41.30ms | 0.129s | 48 | 14.99ms | — |
| hf-019 | 44.50ms | 0.147s | 42 | 15.39ms | — |
| hf-020 | 34.40ms | 0.121s | 41 | 14.72ms | — |
| hf-021 | 41.49ms | 0.070s | 17 | 14.38ms | — |
| hf-022 | 49.40ms | 0.513s | 156 | 51.16ms | — |
| hf-023 | 72.79ms | 0.181s | 46 | 16.02ms | — |
| hf-024 | 68.03ms | 0.143s | 40 | 15.72ms | — |
| hf-025 | 34.79ms | 0.136s | 43 | 15.00ms | — |
| hf-026 | 47.89ms | 0.136s | 42 | 14.90ms | — |
| hf-027 | 42.43ms | 0.130s | 41 | 15.13ms | — |
| hf-028 | 40.35ms | 0.143s | 48 | 15.06ms | — |
| hf-029 | 36.67ms | 0.153s | 42 | 15.12ms | — |
| hf-030 | 35.83ms | 0.138s | 50 | 15.11ms | — |
| hf-031 | 37.33ms | 0.125s | 23 | 15.03ms | — |
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