Round score
0.719
+71.90%
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 | +72.77% | 0.713s | 0.194s | 42 | — |
| hf-001 | +74.03% | 0.732s | 0.190s | 42 | — |
| hf-002 | +74.69% | 0.737s | 0.187s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 92.21ms | 0.194s | 42 | 18.36ms | — |
| hf-001 | 88.21ms | 0.190s | 42 | 18.53ms | — |
| hf-002 | 67.30ms | 0.187s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 42 |
| — |
| hf-003 | +78.87% | 0.803s | 0.170s | 47 | — |
| hf-004 | — | — | — | 28 | candidate output below tolerance |
| hf-006 | +71.63% | 0.468s | 0.133s | 26 | — |
| hf-005 | — | — | — | 28 | candidate output below tolerance |
| hf-007 | +66.08% | 0.597s | 0.203s | 42 | — |
| hf-008 | +64.09% | 0.723s | 0.260s | 48 | — |
| hf-009 | +74.89% | 0.716s | 0.180s | 44 | — |
| hf-010 | +70.99% | 0.707s | 0.205s | 40 | — |
| hf-011 | +69.60% | 0.712s | 0.216s | 41 | — |
| hf-012 | +73.58% | 0.729s | 0.193s | 42 | — |
| hf-013 | +73.35% | 0.734s | 0.196s | 42 | — |
| hf-014 | +77.22% | 0.738s | 0.168s | 42 | — |
| hf-015 | +79.57% | 0.765s | 0.156s | 45 | — |
| hf-017 | — | — | — | 28 | candidate output below tolerance |
| hf-016 | +79.98% | 0.991s | 0.198s | 61 | — |
| hf-018 | +71.73% | 0.898s | 0.254s | 54 | — |
| hf-019 | +70.01% | 0.698s | 0.209s | 42 | — |
| hf-020 | +75.68% | 0.736s | 0.179s | 42 | — |
| hf-021 | +79.83% | 0.752s | 0.152s | 44 | — |
| hf-022 | +72.08% | 0.735s | 0.205s | 42 | — |
| hf-023 | +64.62% | 0.717s | 0.254s | 40 | — |
| hf-024 | +75.85% | 0.750s | 0.181s | 43 | — |
| hf-025 | +72.40% | 0.732s | 0.202s | 42 | — |
| hf-027 | +70.28% | 0.503s | 0.149s | 26 | — |
| hf-026 | +71.68% | 0.735s | 0.208s | 42 | — |
| hf-028 | +66.89% | 0.749s | 0.248s | 43 | — |
| hf-029 | +72.19% | 0.676s | 0.188s | 42 | — |
| hf-030 | +62.77% | 0.659s | 0.245s | 47 | — |
| hf-031 | +65.81% | 0.540s | 0.185s | 43 | — |
| 42 |
| 18.83ms |
| — |
| hf-003 | 67.28ms | 0.170s | 47 | 18.24ms | — |
| hf-004 | 68.48ms | 0.153s | 28 | 18.03ms | — |
| hf-005 | 67.38ms | 0.170s | 28 | 18.31ms | — |
| hf-006 | 64.53ms | 0.133s | 26 | 18.15ms | — |
| hf-007 | 83.30ms | 0.203s | 42 | 17.99ms | — |
| hf-008 | 121.86ms | 0.260s | 48 | 40.44ms | — |
| hf-009 | 94.85ms | 0.180s | 44 | 18.53ms | — |
| hf-010 | 86.04ms | 0.205s | 40 | 18.19ms | — |
| hf-011 | 97.14ms | 0.216s | 41 | 17.93ms | — |
| hf-012 | 90.95ms | 0.210s | 47 | 17.82ms | — |
| hf-013 | 93.45ms | 0.196s | 42 | 18.33ms | — |
| hf-014 | 82.78ms | 0.168s | 42 | 18.46ms | — |
| hf-015 | 54.98ms | 0.156s | 45 | 18.34ms | — |
| hf-016 | 63.30ms | 0.198s | 61 | 19.18ms | — |
| hf-017 | 78.56ms | 0.181s | 28 | 18.27ms | — |
| hf-018 | 76.39ms | 0.254s | 54 | 41.88ms | — |
| hf-019 | 91.18ms | 0.209s | 42 | 19.23ms | — |
| hf-020 | 92.07ms | 0.179s | 42 | 18.40ms | — |
| hf-021 | 66.57ms | 0.152s | 44 | 18.43ms | — |
| hf-022 | 85.52ms | 0.249s | 52 | 43.97ms | — |
| hf-023 | 81.42ms | 0.254s | 40 | 38.08ms | — |
| hf-024 | 96.35ms | 0.181s | 43 | 18.16ms | — |
| hf-025 | 83.33ms | 0.202s | 51 | 18.45ms | — |
| hf-026 | 89.42ms | 0.208s | 42 | 18.48ms | — |
| hf-027 | 80.87ms | 0.215s | 67 | 18.15ms | — |
| hf-028 | 90.78ms | 0.248s | 43 | 39.03ms | — |
| hf-029 | 86.55ms | 0.188s | 42 | 18.33ms | — |
| hf-030 | 103.69ms | 0.245s | 47 | 39.00ms | — |
| hf-031 | 83.45ms | 0.201s | 44 | 18.11ms | — |
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