Round score
0.6906
+69.06%
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 | +69.16% | 0.591s | 0.182s | 34 | — |
| hf-001 | — | — | — | 29 | candidate output below tolerance |
| hf-002 | +65.41% | 0.794s | 0.275s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 81.84ms | 0.302s | 64 | 70.12ms | — |
| hf-001 | 101.89ms | 0.185s | 29 | 16.96ms | — |
| hf-002 | 88.53ms | 0.307s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 48 |
| — |
| hf-003 | +68.97% | 0.677s | 0.210s | 40 | — |
| hf-004 | +60.43% | 0.741s | 0.293s | 42 | — |
| hf-005 | +70.59% | 0.715s | 0.210s | 43 | — |
| hf-006 | +59.60% | 0.757s | 0.306s | 48 | — |
| hf-007 | +55.75% | 0.734s | 0.325s | 43 | — |
| hf-008 | — | — | — | 23 | candidate output below tolerance |
| hf-009 | +74.69% | 0.777s | 0.197s | 43 | — |
| hf-010 | +74.19% | 0.774s | 0.200s | 42 | — |
| hf-011 | +77.78% | 0.774s | 0.172s | 43 | — |
| hf-012 | +76.90% | 0.749s | 0.173s | 42 | — |
| hf-013 | +74.52% | 0.731s | 0.186s | 40 | — |
| hf-015 | +59.33% | 0.376s | 0.153s | 20 | — |
| hf-017 | +55.17% | 0.411s | 0.184s | 23 | — |
| hf-016 | +59.13% | 0.701s | 0.287s | 40 | — |
| hf-014 | — | — | — | 28 | candidate output below tolerance |
| hf-019 | +63.61% | 0.526s | 0.191s | 29 | — |
| hf-018 | +71.95% | 0.779s | 0.219s | 46 | — |
| hf-021 | +71.36% | 0.526s | 0.151s | 25 | — |
| hf-020 | +75.47% | 0.780s | 0.191s | 43 | — |
| hf-022 | — | — | — | 44 | candidate output below tolerance |
| hf-023 | +71.09% | 0.755s | 0.218s | 42 | — |
| hf-025 | +76.63% | 0.746s | 0.174s | 42 | — |
| hf-024 | +76.17% | 0.787s | 0.188s | 46 | — |
| hf-026 | +77.32% | 0.765s | 0.174s | 43 | — |
| hf-027 | +75.53% | 0.777s | 0.190s | 43 | — |
| hf-028 | +78.26% | 0.752s | 0.163s | 41 | — |
| hf-029 | — | — | — | 28 | candidate output below tolerance |
| hf-030 | +67.00% | 0.533s | 0.176s | 34 | — |
| hf-031 | +56.95% | 0.533s | 0.229s | 42 | — |
| 59 |
| 68.99ms |
| — |
| hf-003 | 89.64ms | 0.210s | 40 | 19.37ms | — |
| hf-004 | 89.00ms | 0.293s | 42 | 69.08ms | — |
| hf-005 | 93.23ms | 0.210s | 43 | 17.59ms | — |
| hf-006 | 128.44ms | 0.306s | 48 | 81.69ms | — |
| hf-007 | 105.77ms | 0.325s | 43 | 68.43ms | — |
| hf-008 | 91.25ms | 0.195s | 23 | 18.91ms | — |
| hf-009 | 80.95ms | 0.197s | 43 | 16.98ms | — |
| hf-010 | 82.99ms | 0.200s | 42 | 17.70ms | — |
| hf-011 | 72.41ms | 0.172s | 43 | 17.34ms | — |
| hf-012 | 72.91ms | 0.173s | 42 | 17.47ms | — |
| hf-013 | 85.38ms | 0.186s | 40 | 17.78ms | — |
| hf-014 | 63.26ms | 0.146s | 28 | 16.99ms | — |
| hf-015 | 101.50ms | 0.202s | 41 | 17.79ms | — |
| hf-016 | 105.31ms | 0.287s | 40 | 72.96ms | — |
| hf-017 | 86.62ms | 0.184s | 23 | 16.97ms | — |
| hf-018 | 101.39ms | 0.219s | 46 | 17.33ms | — |
| hf-019 | 89.46ms | 0.191s | 29 | 18.35ms | — |
| hf-020 | 75.84ms | 0.191s | 44 | 18.07ms | — |
| hf-021 | 64.02ms | 0.300s | 54 | 70.88ms | — |
| hf-022 | 100.06ms | 0.301s | 44 | 72.69ms | — |
| hf-023 | 101.48ms | 0.218s | 42 | 17.20ms | — |
| hf-024 | 87.50ms | 0.188s | 46 | 17.40ms | — |
| hf-025 | 75.15ms | 0.174s | 42 | 17.08ms | — |
| hf-026 | 73.58ms | 0.174s | 43 | 17.41ms | — |
| hf-027 | 74.36ms | 0.190s | 43 | 19.02ms | — |
| hf-028 | 63.86ms | 0.179s | 43 | 17.04ms | — |
| hf-029 | 73.30ms | 0.157s | 28 | 17.71ms | — |
| hf-030 | 75.14ms | 0.298s | 64 | 71.56ms | — |
| hf-031 | 98.46ms | 0.229s | 42 | 16.73ms | — |
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