Round score
0.1702
+17.02%
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 | +21.51% | 0.489s | 0.383s | 28 | — |
| hf-001 | +16.87% | 0.871s | 0.724s | 52 | — |
| hf-002 | +9.99% | 0.788s | 0.709s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 61.28ms | 0.766s | 57 | 55.59ms | — |
| hf-001 | 66.37ms | 0.724s | 52 | 55.59ms | — |
| hf-002 | 62.25ms | 0.709s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 51 |
| — |
| hf-003 | +19.94% | 0.730s | 0.584s | 42 | — |
| hf-004 | +20.11% | 0.740s | 0.591s | 43 | — |
| hf-005 | +19.72% | 0.748s | 0.600s | 43 | — |
| hf-006 | +19.39% | 0.706s | 0.569s | 40 | — |
| hf-007 | +18.31% | 0.738s | 0.603s | 43 | — |
| hf-009 | +10.69% | 0.782s | 0.699s | 50 | — |
| hf-008 | +17.07% | 1.123s | 0.931s | 68 | — |
| hf-010 | +19.89% | 0.755s | 0.605s | 44 | — |
| hf-011 | +21.74% | 0.715s | 0.560s | 40 | — |
| hf-012 | +25.06% | 0.935s | 0.701s | 50 | — |
| hf-014 | +26.01% | 0.758s | 0.561s | 41 | — |
| hf-013 | — | — | — | 27 | candidate output below tolerance |
| hf-015 | +26.56% | 0.751s | 0.552s | 40 | — |
| hf-016 | +24.67% | 0.783s | 0.590s | 43 | — |
| hf-017 | +17.13% | 0.803s | 0.666s | 45 | — |
| hf-018 | — | — | — | 28 | candidate output below tolerance |
| hf-020 | +21.77% | 0.707s | 0.553s | 40 | — |
| hf-019 | +14.80% | 1.013s | 0.863s | 61 | — |
| hf-021 | +16.97% | 0.672s | 0.558s | 40 | — |
| hf-022 | +7.80% | 0.599s | 0.552s | 40 | — |
| hf-023 | +14.60% | 0.779s | 0.666s | 47 | — |
| hf-024 | +21.16% | 0.728s | 0.574s | 42 | — |
| hf-025 | — | — | — | 42 | candidate output below tolerance |
| hf-026 | — | — | — | 23 | candidate output below tolerance |
| hf-027 | — | — | — | 34 | candidate output below tolerance |
| hf-029 | +12.51% | 0.645s | 0.564s | 41 | — |
| hf-028 | +14.30% | 0.804s | 0.689s | 49 | — |
| hf-030 | +6.94% | 0.634s | 0.590s | 47 | — |
| hf-031 | +7.44% | 0.517s | 0.478s | 40 | — |
| 51 |
| 55.65ms |
| — |
| hf-003 | 70.44ms | 0.584s | 42 | 55.65ms | — |
| hf-004 | 69.75ms | 0.591s | 43 | 56.54ms | — |
| hf-005 | 66.81ms | 0.600s | 43 | 56.54ms | — |
| hf-006 | 65.61ms | 0.569s | 40 | 55.88ms | — |
| hf-007 | 83.02ms | 0.603s | 43 | 55.11ms | — |
| hf-008 | 69.81ms | 0.931s | 68 | 55.26ms | — |
| hf-009 | 64.84ms | 0.699s | 50 | 55.15ms | — |
| hf-010 | 70.05ms | 0.768s | 56 | 55.35ms | — |
| hf-011 | 66.12ms | 0.560s | 40 | 55.19ms | — |
| hf-012 | 71.35ms | 0.723s | 52 | 55.13ms | — |
| hf-013 | 72.88ms | 0.386s | 27 | 55.17ms | — |
| hf-014 | 64.99ms | 0.561s | 41 | 56.04ms | — |
| hf-015 | 64.77ms | 0.563s | 41 | 55.38ms | — |
| hf-016 | 71.35ms | 0.590s | 43 | 55.27ms | — |
| hf-017 | 67.28ms | 0.666s | 45 | 55.50ms | — |
| hf-018 | 62.43ms | 0.400s | 28 | 54.96ms | — |
| hf-019 | 80.36ms | 0.863s | 61 | 55.47ms | — |
| hf-020 | 67.97ms | 0.553s | 40 | 55.30ms | — |
| hf-021 | 71.09ms | 0.558s | 40 | 55.25ms | — |
| hf-022 | 66.30ms | 0.552s | 40 | 55.29ms | — |
| hf-023 | 62.62ms | 0.668s | 48 | 55.27ms | — |
| hf-024 | 66.49ms | 0.574s | 42 | 56.15ms | — |
| hf-025 | 72.66ms | 0.582s | 42 | 55.27ms | — |
| hf-026 | 68.10ms | 0.341s | 23 | 54.79ms | — |
| hf-027 | 65.91ms | 0.489s | 34 | 55.26ms | — |
| hf-028 | 73.33ms | 0.689s | 49 | 56.18ms | — |
| hf-029 | 67.90ms | 0.564s | 41 | 56.20ms | — |
| hf-030 | 67.23ms | 0.590s | 47 | 55.75ms | — |
| hf-031 | 68.71ms | 0.478s | 40 | 10.96ms | — |
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