Round score
0.7339
+73.39%
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 | +57.06% | 0.591s | 0.254s | 34 | — |
| hf-001 | +76.53% | 0.713s | 0.167s | 41 | — |
| hf-002 | +67.93% | 0.773s | 0.248s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 114.24ms | 0.288s | 46 | 43.02ms | — |
| hf-001 | 88.52ms | 0.167s | 41 | 16.35ms | — |
| hf-002 | 124.77ms | 0.248s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 46 |
| — |
| hf-003 | +72.45% | 0.709s | 0.195s | 43 | — |
| hf-004 | +75.86% | 0.741s | 0.179s | 42 | — |
| hf-005 | +76.88% | 0.715s | 0.165s | 43 | — |
| hf-006 | +70.77% | 0.735s | 0.215s | 46 | — |
| hf-007 | +65.34% | 0.734s | 0.254s | 43 | — |
| hf-008 | +72.84% | 0.730s | 0.198s | 43 | — |
| hf-009 | +75.11% | 0.744s | 0.185s | 40 | — |
| hf-010 | +79.08% | 0.774s | 0.162s | 42 | — |
| hf-011 | +78.82% | 0.774s | 0.164s | 43 | — |
| hf-012 | +74.51% | 0.771s | 0.197s | 44 | — |
| hf-013 | +75.79% | 0.731s | 0.177s | 40 | — |
| hf-015 | +62.96% | 0.376s | 0.139s | 20 | — |
| hf-017 | +67.73% | 0.466s | 0.150s | 25 | — |
| hf-016 | +72.68% | 0.712s | 0.195s | 41 | — |
| hf-014 | — | — | — | 31 | candidate output below tolerance |
| hf-019 | +72.38% | 0.526s | 0.145s | 29 | — |
| hf-018 | +72.93% | 0.779s | 0.211s | 46 | — |
| hf-021 | +77.44% | 0.526s | 0.119s | 25 | — |
| hf-020 | — | — | — | 22 | candidate output below tolerance |
| hf-022 | — | — | — | 24 | candidate output below tolerance |
| hf-023 | +74.96% | 0.755s | 0.189s | 42 | — |
| hf-025 | +74.20% | 0.746s | 0.192s | 42 | — |
| hf-024 | +80.97% | 0.787s | 0.150s | 46 | — |
| hf-026 | +78.81% | 0.765s | 0.162s | 43 | — |
| hf-027 | +73.85% | 0.734s | 0.192s | 39 | — |
| hf-028 | +76.54% | 0.752s | 0.176s | 41 | — |
| hf-029 | +76.57% | 0.680s | 0.159s | 41 | — |
| hf-030 | +59.01% | 0.533s | 0.218s | 34 | — |
| hf-031 | +58.20% | 0.533s | 0.223s | 42 | — |
| 46 |
| 42.08ms |
| — |
| hf-003 | 85.04ms | 0.195s | 43 | 17.51ms | — |
| hf-004 | 69.99ms | 0.194s | 45 | 19.06ms | — |
| hf-005 | 69.67ms | 0.165s | 43 | 18.25ms | — |
| hf-006 | 103.11ms | 0.215s | 46 | 18.23ms | — |
| hf-007 | 85.75ms | 0.254s | 43 | 41.98ms | — |
| hf-008 | 87.51ms | 0.198s | 43 | 17.45ms | — |
| hf-009 | 74.90ms | 0.185s | 40 | 17.35ms | — |
| hf-010 | 67.14ms | 0.177s | 46 | 16.93ms | — |
| hf-011 | 53.78ms | 0.195s | 55 | 18.38ms | — |
| hf-012 | 69.54ms | 0.197s | 44 | 17.63ms | — |
| hf-013 | 80.92ms | 0.177s | 40 | 17.99ms | — |
| hf-014 | 54.29ms | 0.148s | 31 | 17.62ms | — |
| hf-015 | 90.62ms | 0.186s | 46 | 17.57ms | — |
| hf-016 | 82.77ms | 0.210s | 48 | 17.83ms | — |
| hf-017 | 70.86ms | 0.150s | 25 | 17.09ms | — |
| hf-018 | 84.04ms | 0.211s | 46 | 17.47ms | — |
| hf-019 | 81.05ms | 0.145s | 29 | 17.48ms | — |
| hf-020 | 63.81ms | 0.160s | 22 | 17.54ms | — |
| hf-021 | 54.42ms | 0.119s | 25 | 17.70ms | — |
| hf-022 | 62.97ms | 0.143s | 24 | 16.98ms | — |
| hf-023 | 62.53ms | 0.189s | 42 | 17.12ms | — |
| hf-024 | 54.88ms | 0.150s | 46 | 18.07ms | — |
| hf-025 | 64.35ms | 0.192s | 43 | 17.02ms | — |
| hf-026 | 65.58ms | 0.162s | 43 | 17.87ms | — |
| hf-027 | 65.16ms | 0.192s | 39 | 21.83ms | — |
| hf-028 | 64.97ms | 0.176s | 41 | 17.75ms | — |
| hf-029 | 64.55ms | 0.159s | 41 | 17.52ms | — |
| hf-030 | 106.61ms | 0.264s | 46 | 45.60ms | — |
| hf-031 | 97.01ms | 0.223s | 42 | 17.63ms | — |
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