Round score
0.7241
+72.41%
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-000 | +72.66% | 0.713s | 0.195s | 42 | — |
| hf-001 | +73.70% | 0.732s | 0.192s | 42 | — |
| hf-002 | +77.56% | 0.737s | 0.165s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 83.55ms | 0.195s | 42 | 17.47ms | — |
| hf-001 | 80.54ms | 0.192s | 42 | 17.80ms | — |
| hf-002 | 69.29ms | 0.165s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 42 |
| — |
| hf-003 | +79.83% | 0.803s | 0.162s | 47 | — |
| hf-004 | — | — | — | 28 | candidate output below tolerance |
| hf-006 | +69.44% | 0.468s | 0.143s | 26 | — |
| hf-005 | — | — | — | 28 | candidate output below tolerance |
| hf-007 | +65.04% | 0.597s | 0.209s | 42 | — |
| hf-008 | +59.53% | 0.723s | 0.293s | 48 | — |
| hf-009 | +73.84% | 0.716s | 0.187s | 44 | — |
| hf-010 | +70.14% | 0.707s | 0.211s | 40 | — |
| hf-011 | +70.25% | 0.712s | 0.212s | 41 | — |
| hf-012 | +73.14% | 0.729s | 0.196s | 42 | — |
| hf-013 | +75.82% | 0.734s | 0.177s | 42 | — |
| hf-014 | +78.90% | 0.738s | 0.156s | 42 | — |
| hf-015 | +77.20% | 0.765s | 0.174s | 45 | — |
| hf-017 | — | — | — | 28 | candidate output below tolerance |
| hf-016 | +75.45% | 0.991s | 0.243s | 61 | — |
| hf-018 | +75.74% | 0.898s | 0.218s | 54 | — |
| hf-019 | +69.05% | 0.698s | 0.216s | 42 | — |
| hf-020 | +73.73% | 0.736s | 0.193s | 42 | — |
| hf-021 | +77.99% | 0.752s | 0.165s | 44 | — |
| hf-022 | +71.48% | 0.735s | 0.210s | 42 | — |
| hf-023 | +59.35% | 0.717s | 0.292s | 40 | — |
| hf-024 | +73.04% | 0.750s | 0.202s | 43 | — |
| hf-025 | +73.97% | 0.732s | 0.191s | 42 | — |
| hf-027 | +72.21% | 0.503s | 0.140s | 26 | — |
| hf-026 | +72.61% | 0.735s | 0.201s | 42 | — |
| hf-028 | +65.84% | 0.749s | 0.256s | 43 | — |
| hf-029 | +69.88% | 0.676s | 0.204s | 42 | — |
| hf-030 | +55.99% | 0.659s | 0.290s | 47 | — |
| hf-031 | +62.92% | 0.540s | 0.200s | 43 | — |
| 42 |
| 18.33ms |
| — |
| hf-003 | 65.87ms | 0.162s | 47 | 17.30ms | — |
| hf-004 | 66.98ms | 0.147s | 28 | 17.73ms | — |
| hf-005 | 66.65ms | 0.163s | 28 | 17.39ms | — |
| hf-006 | 63.66ms | 0.143s | 26 | 17.56ms | — |
| hf-007 | 82.71ms | 0.209s | 42 | 17.01ms | — |
| hf-008 | 118.78ms | 0.293s | 48 | 38.63ms | — |
| hf-009 | 92.52ms | 0.187s | 44 | 16.91ms | — |
| hf-010 | 84.74ms | 0.211s | 40 | 17.18ms | — |
| hf-011 | 100.77ms | 0.212s | 41 | 17.20ms | — |
| hf-012 | 84.95ms | 0.196s | 47 | 17.12ms | — |
| hf-013 | 81.10ms | 0.177s | 42 | 17.58ms | — |
| hf-014 | 75.20ms | 0.156s | 42 | 17.65ms | — |
| hf-015 | 63.62ms | 0.174s | 45 | 17.47ms | — |
| hf-016 | 63.14ms | 0.243s | 61 | 39.02ms | — |
| hf-017 | 85.58ms | 0.181s | 28 | 17.49ms | — |
| hf-018 | 75.38ms | 0.218s | 54 | 17.48ms | — |
| hf-019 | 89.66ms | 0.216s | 42 | 17.59ms | — |
| hf-020 | 97.51ms | 0.193s | 42 | 17.54ms | — |
| hf-021 | 69.43ms | 0.165s | 44 | 17.35ms | — |
| hf-022 | 83.53ms | 0.251s | 52 | 40.97ms | — |
| hf-023 | 85.15ms | 0.292s | 40 | 49.47ms | — |
| hf-024 | 91.70ms | 0.202s | 43 | 17.00ms | — |
| hf-025 | 94.05ms | 0.206s | 51 | 17.53ms | — |
| hf-026 | 89.38ms | 0.201s | 42 | 17.47ms | — |
| hf-027 | 75.58ms | 0.217s | 67 | 17.46ms | — |
| hf-028 | 90.74ms | 0.256s | 43 | 38.56ms | — |
| hf-029 | 93.18ms | 0.204s | 42 | 17.10ms | — |
| hf-030 | 115.75ms | 0.290s | 47 | 38.75ms | — |
| hf-031 | 89.85ms | 0.200s | 44 | 16.95ms | — |
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