Round score
0.6211
+62.11%
Prompts scored
29 / 32
median_e2e_speedup
Below tolerance
3
answered too little to score
Zeroed total
3
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 | +69.57% | 0.700s | 0.213s | 42 | — |
| hf-002 | +71.81% | 0.657s | 0.185s | 37 | — |
| hf-000 | — | — | — |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 102.03ms | 0.352s | 46 | 39.74ms | — |
| hf-001 | 87.98ms | 0.213s | 42 | 16.52ms | — |
| hf-002 | 76.75ms | 0.215s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 46 |
| candidate output below tolerance |
| hf-003 | +70.19% | 0.710s | 0.212s | 43 | — |
| hf-005 | +56.22% | 0.374s | 0.164s | 24 | — |
| hf-004 | +72.38% | 0.773s | 0.214s | 48 | — |
| hf-007 | +55.51% | 0.372s | 0.165s | 22 | — |
| hf-006 | +57.85% | 0.716s | 0.302s | 43 | — |
| hf-008 | +60.54% | 0.860s | 0.339s | 53 | — |
| hf-009 | +70.32% | 0.726s | 0.215s | 43 | — |
| hf-010 | +58.31% | 0.724s | 0.302s | 43 | — |
| hf-011 | +69.42% | 0.683s | 0.209s | 43 | — |
| hf-012 | +51.76% | 0.591s | 0.285s | 41 | — |
| hf-013 | +59.87% | 0.716s | 0.287s | 42 | — |
| hf-014 | +56.64% | 0.705s | 0.306s | 41 | — |
| hf-015 | +60.54% | 0.736s | 0.290s | 41 | — |
| hf-016 | +71.77% | 0.704s | 0.199s | 42 | — |
| hf-018 | +69.22% | 0.693s | 0.213s | 43 | — |
| hf-017 | +62.80% | 0.947s | 0.352s | 61 | — |
| hf-019 | +65.81% | 0.612s | 0.209s | 42 | — |
| hf-021 | +64.20% | 0.460s | 0.165s | 28 | — |
| hf-020 | — | — | — | 20 | candidate output below tolerance |
| hf-022 | +59.77% | 0.367s | 0.148s | 23 | — |
| hf-024 | +65.52% | 0.414s | 0.143s | 28 | — |
| hf-023 | — | — | — | 42 | candidate output below tolerance |
| hf-026 | +63.79% | 0.529s | 0.192s | 28 | — |
| hf-025 | +73.67% | 0.784s | 0.207s | 46 | — |
| hf-027 | +62.05% | 0.716s | 0.272s | 43 | — |
| hf-029 | +38.51% | 0.302s | 0.185s | 17 | — |
| hf-028 | +54.74% | 0.612s | 0.277s | 40 | — |
| hf-030 | +64.55% | 0.615s | 0.218s | 47 | — |
| hf-031 | +62.17% | 0.762s | 0.288s | 68 | — |
| 45 |
| 15.87ms |
| — |
| hf-003 | 88.46ms | 0.212s | 43 | 15.97ms | — |
| hf-004 | 75.45ms | 0.214s | 48 | 15.88ms | — |
| hf-005 | 86.76ms | 0.317s | 49 | 67.09ms | — |
| hf-006 | 117.75ms | 0.302s | 43 | 64.65ms | — |
| hf-007 | 106.50ms | 0.252s | 43 | 40.10ms | — |
| hf-008 | 88.70ms | 0.339s | 53 | 44.01ms | — |
| hf-009 | 91.17ms | 0.280s | 46 | 64.20ms | — |
| hf-010 | 115.46ms | 0.302s | 43 | 40.55ms | — |
| hf-011 | 85.79ms | 0.209s | 43 | 16.39ms | — |
| hf-012 | 122.56ms | 0.285s | 41 | 66.68ms | — |
| hf-013 | 110.80ms | 0.287s | 42 | 70.96ms | — |
| hf-014 | 133.97ms | 0.306s | 41 | 64.50ms | — |
| hf-015 | 106.30ms | 0.307s | 46 | 42.49ms | — |
| hf-016 | 90.20ms | 0.199s | 42 | 16.47ms | — |
| hf-017 | 99.07ms | 0.352s | 61 | 40.23ms | — |
| hf-018 | 88.38ms | 0.213s | 45 | 16.91ms | — |
| hf-019 | 85.63ms | 0.209s | 42 | 16.27ms | — |
| hf-020 | 106.52ms | 0.196s | 20 | 15.64ms | — |
| hf-021 | 72.05ms | 0.165s | 28 | 16.02ms | — |
| hf-022 | 55.69ms | 0.148s | 23 | 15.67ms | — |
| hf-023 | 65.53ms | 0.173s | 42 | 15.87ms | — |
| hf-024 | 65.10ms | 0.173s | 42 | 15.98ms | — |
| hf-025 | 67.88ms | 0.207s | 46 | 16.48ms | — |
| hf-026 | 83.14ms | 0.192s | 28 | 16.11ms | — |
| hf-027 | 81.02ms | 0.295s | 47 | 65.69ms | — |
| hf-028 | 95.31ms | 0.322s | 45 | 71.99ms | — |
| hf-029 | 122.06ms | 0.317s | 43 | 37.55ms | — |
| hf-030 | 79.56ms | 0.218s | 47 | 16.23ms | — |
| hf-031 | 88.82ms | 0.288s | 68 | 16.20ms | — |
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