Round score
0.779
+77.90%
Prompts scored
26 / 32
median_e2e_speedup
Below tolerance
6
answered too little to score
Zeroed total
6
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 | +76.09% | 0.712s | 0.170s | 42 | — |
| hf-001 | +80.86% | 0.760s | 0.145s | 45 | — |
| hf-003 | +84.49% | 0.755s | 0.117s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 50.21ms | 0.170s | 42 | 15.10ms | — |
| hf-001 | 53.82ms | 0.145s | 45 | 15.84ms | — |
| hf-002 | 62.88ms | 0.172s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 44 |
| — |
| hf-002 | — | — | — | 51 | candidate output below tolerance |
| hf-004 | +81.53% | 0.730s | 0.135s | 42 | — |
| hf-005 | +78.69% | 0.753s | 0.161s | 47 | — |
| hf-006 | +80.81% | 0.694s | 0.133s | 42 | — |
| hf-007 | — | — | — | 28 | candidate output below tolerance |
| hf-009 | — | — | — | 28 | candidate output below tolerance |
| hf-008 | +72.24% | 0.942s | 0.262s | 63 | — |
| hf-010 | — | — | — | 16 | candidate output below tolerance |
| hf-012 | +71.87% | 0.580s | 0.163s | 37 | — |
| hf-011 | +79.02% | 0.903s | 0.189s | 52 | — |
| hf-013 | +70.73% | 0.497s | 0.145s | 28 | — |
| hf-014 | +82.07% | 0.739s | 0.133s | 42 | — |
| hf-015 | +78.26% | 0.717s | 0.156s | 41 | — |
| hf-016 | +83.67% | 0.739s | 0.121s | 43 | — |
| hf-017 | +78.61% | 0.742s | 0.159s | 43 | — |
| hf-018 | +79.41% | 0.704s | 0.145s | 42 | — |
| hf-019 | +79.93% | 0.872s | 0.175s | 54 | — |
| hf-020 | +80.00% | 0.662s | 0.132s | 41 | — |
| hf-021 | — | — | — | 42 | candidate output below tolerance |
| hf-023 | — | — | — | 26 | candidate output below tolerance |
| hf-022 | +77.38% | 0.653s | 0.148s | 41 | — |
| hf-024 | +79.21% | 0.701s | 0.146s | 46 | — |
| hf-026 | +69.41% | 0.359s | 0.110s | 23 | — |
| hf-025 | +80.20% | 0.591s | 0.117s | 42 | — |
| hf-029 | +65.02% | 0.325s | 0.114s | 20 | — |
| hf-028 | +75.08% | 0.708s | 0.176s | 41 | — |
| hf-027 | +79.08% | 0.812s | 0.170s | 54 | — |
| hf-030 | +77.54% | 0.586s | 0.132s | 41 | — |
| hf-031 | +77.03% | 0.297s | 0.068s | 20 | — |
| 51 |
| 16.89ms |
| — |
| hf-003 | 28.05ms | 0.117s | 44 | 15.30ms | — |
| hf-004 | 44.35ms | 0.135s | 42 | 15.25ms | — |
| hf-005 | 54.14ms | 0.161s | 48 | 15.46ms | — |
| hf-006 | 29.00ms | 0.133s | 42 | 15.11ms | — |
| hf-007 | 37.65ms | 0.128s | 28 | 15.21ms | — |
| hf-008 | 62.86ms | 0.268s | 75 | 43.45ms | — |
| hf-009 | 67.18ms | 0.143s | 28 | 15.52ms | — |
| hf-010 | 62.68ms | 0.094s | 16 | 16.04ms | — |
| hf-011 | 53.08ms | 0.189s | 52 | 16.07ms | — |
| hf-012 | 29.57ms | 0.251s | 57 | 42.95ms | — |
| hf-013 | 84.03ms | 0.145s | 28 | 16.05ms | — |
| hf-014 | 42.27ms | 0.133s | 47 | 15.28ms | — |
| hf-015 | 63.44ms | 0.156s | 41 | 16.48ms | — |
| hf-016 | 31.96ms | 0.121s | 43 | 15.29ms | — |
| hf-017 | 53.10ms | 0.159s | 43 | 15.54ms | — |
| hf-018 | 54.04ms | 0.160s | 45 | 15.58ms | — |
| hf-019 | 53.54ms | 0.175s | 54 | 16.02ms | — |
| hf-020 | 28.35ms | 0.132s | 41 | 15.33ms | — |
| hf-021 | 63.26ms | 0.173s | 42 | 19.09ms | — |
| hf-022 | 29.05ms | 0.148s | 41 | 15.43ms | — |
| hf-023 | 62.65ms | 0.128s | 26 | 17.99ms | — |
| hf-024 | 40.24ms | 0.146s | 46 | 15.60ms | — |
| hf-025 | 27.77ms | 0.117s | 42 | 15.15ms | — |
| hf-026 | 63.24ms | 0.204s | 52 | 17.85ms | — |
| hf-027 | 74.49ms | 0.185s | 63 | 17.84ms | — |
| hf-028 | 66.25ms | 0.176s | 41 | 19.01ms | — |
| hf-029 | 63.38ms | 0.174s | 45 | 18.16ms | — |
| hf-030 | 55.35ms | 0.147s | 46 | 15.66ms | — |
| hf-031 | 38.08ms | 0.113s | 43 | 15.26ms | — |
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