Round score
0.6768
+67.68%
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-001 | +56.63% | 0.700s | 0.303s | 42 | — |
| hf-002 | +62.58% | 0.657s | 0.246s | 37 | — |
| hf-000 | — | — | — |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 99.77ms | 0.298s | 47 | 39.94ms | — |
| hf-001 | 98.23ms | 0.303s | 43 | 41.13ms | — |
| hf-002 | 86.03ms | 0.284s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 47 |
| candidate output below tolerance |
| hf-003 | +75.91% | 0.700s | 0.169s | 42 | — |
| hf-005 | +64.48% | 0.374s | 0.133s | 24 | — |
| hf-004 | — | — | — | 18 | candidate output below tolerance |
| hf-007 | +61.50% | 0.372s | 0.143s | 22 | — |
| hf-006 | +71.40% | 0.716s | 0.205s | 43 | — |
| hf-008 | — | — | — | 29 | candidate output below tolerance |
| hf-009 | +76.51% | 0.726s | 0.170s | 43 | — |
| hf-010 | +73.17% | 0.704s | 0.189s | 41 | — |
| hf-011 | +73.07% | 0.683s | 0.184s | 43 | — |
| hf-012 | +67.89% | 0.591s | 0.190s | 41 | — |
| hf-013 | +74.84% | 0.716s | 0.180s | 42 | — |
| hf-014 | +73.25% | 0.705s | 0.189s | 41 | — |
| hf-015 | +75.69% | 0.736s | 0.179s | 41 | — |
| hf-016 | +74.00% | 0.704s | 0.183s | 42 | — |
| hf-018 | — | — | — | 23 | candidate output below tolerance |
| hf-017 | — | — | — | 51 | candidate output below tolerance |
| hf-019 | +68.98% | 0.612s | 0.190s | 42 | — |
| hf-021 | +67.46% | 0.460s | 0.150s | 28 | — |
| hf-020 | +71.92% | 0.726s | 0.204s | 47 | — |
| hf-022 | +62.12% | 0.367s | 0.139s | 23 | — |
| hf-024 | +51.82% | 0.414s | 0.200s | 28 | — |
| hf-023 | +74.50% | 0.727s | 0.185s | 47 | — |
| hf-026 | +59.60% | 0.529s | 0.214s | 28 | — |
| hf-025 | +70.51% | 0.742s | 0.219s | 42 | — |
| hf-027 | +73.84% | 0.705s | 0.184s | 42 | — |
| hf-029 | +36.18% | 0.302s | 0.192s | 17 | — |
| hf-028 | +68.25% | 0.612s | 0.194s | 40 | — |
| hf-030 | +59.80% | 0.615s | 0.247s | 47 | — |
| hf-031 | +64.65% | 0.762s | 0.270s | 68 | — |
| 43 |
| 41.95ms |
| — |
| hf-003 | 83.64ms | 0.169s | 42 | 17.34ms | — |
| hf-004 | 55.94ms | 0.106s | 18 | 17.36ms | — |
| hf-005 | 64.32ms | 0.199s | 49 | 17.48ms | — |
| hf-006 | 103.57ms | 0.205s | 43 | 17.67ms | — |
| hf-007 | 91.88ms | 0.193s | 43 | 17.59ms | — |
| hf-008 | 70.55ms | 0.189s | 29 | 17.74ms | — |
| hf-009 | 69.51ms | 0.187s | 46 | 17.74ms | — |
| hf-010 | 88.37ms | 0.189s | 41 | 17.30ms | — |
| hf-011 | 66.82ms | 0.184s | 43 | 17.34ms | — |
| hf-012 | 105.53ms | 0.190s | 41 | 17.49ms | — |
| hf-013 | 78.73ms | 0.180s | 42 | 17.49ms | — |
| hf-014 | 103.77ms | 0.189s | 41 | 17.32ms | — |
| hf-015 | 77.75ms | 0.179s | 41 | 17.38ms | — |
| hf-016 | 65.68ms | 0.183s | 42 | 17.27ms | — |
| hf-017 | 86.27ms | 0.321s | 51 | 46.26ms | — |
| hf-018 | 90.48ms | 0.194s | 23 | 18.00ms | — |
| hf-019 | 71.64ms | 0.190s | 42 | 18.20ms | — |
| hf-020 | 85.00ms | 0.204s | 47 | 17.73ms | — |
| hf-021 | 81.05ms | 0.150s | 28 | 17.80ms | — |
| hf-022 | 55.65ms | 0.247s | 59 | 41.77ms | — |
| hf-023 | 85.54ms | 0.185s | 47 | 17.00ms | — |
| hf-024 | 66.20ms | 0.335s | 45 | 42.47ms | — |
| hf-025 | 99.06ms | 0.219s | 42 | 17.94ms | — |
| hf-026 | 95.90ms | 0.214s | 28 | 17.18ms | — |
| hf-027 | 99.75ms | 0.184s | 42 | 17.13ms | — |
| hf-028 | 76.06ms | 0.211s | 45 | 17.51ms | — |
| hf-029 | 123.34ms | 0.323s | 43 | 50.34ms | — |
| hf-030 | 88.22ms | 0.247s | 47 | 39.61ms | — |
| hf-031 | 86.16ms | 0.270s | 68 | 18.24ms | — |
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