Round score
0.718
+71.80%
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-000 | +70.01% | 0.591s | 0.177s | 34 | — |
| hf-001 | +72.04% | 0.734s | 0.205s | 43 | — |
| hf-002 | — | — | — |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 79.91ms | 0.292s | 58 | 69.90ms | — |
| hf-001 | 92.55ms | 0.205s | 43 | 16.75ms | — |
| hf-002 | 95.26ms | 0.128s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 16 |
| candidate output below tolerance |
| hf-003 | +70.58% | 0.709s | 0.209s | 43 | — |
| hf-004 | — | — | — | 23 | candidate output below tolerance |
| hf-005 | +73.82% | 0.715s | 0.187s | 43 | — |
| hf-006 | +62.35% | 0.757s | 0.285s | 48 | — |
| hf-007 | +60.23% | 0.734s | 0.292s | 43 | — |
| hf-008 | +71.55% | 0.730s | 0.208s | 43 | — |
| hf-009 | +74.44% | 0.777s | 0.199s | 43 | — |
| hf-010 | +74.79% | 0.774s | 0.195s | 42 | — |
| hf-011 | +74.99% | 0.774s | 0.194s | 43 | — |
| hf-012 | +75.06% | 0.749s | 0.187s | 42 | — |
| hf-013 | +73.04% | 0.731s | 0.197s | 40 | — |
| hf-015 | +51.61% | 0.376s | 0.182s | 20 | — |
| hf-017 | +63.03% | 0.466s | 0.172s | 25 | — |
| hf-016 | +58.85% | 0.712s | 0.293s | 41 | — |
| hf-014 | — | — | — | 18 | candidate output below tolerance |
| hf-019 | +71.11% | 0.526s | 0.152s | 29 | — |
| hf-018 | — | — | — | 16 | candidate output below tolerance |
| hf-021 | +75.90% | 0.526s | 0.127s | 25 | — |
| hf-020 | +76.47% | 0.780s | 0.184s | 43 | — |
| hf-022 | +77.21% | 0.944s | 0.215s | 54 | — |
| hf-023 | — | — | — | 18 | candidate output below tolerance |
| hf-025 | +77.36% | 0.746s | 0.169s | 42 | — |
| hf-024 | +79.90% | 0.787s | 0.158s | 46 | — |
| hf-026 | +75.95% | 0.765s | 0.184s | 43 | — |
| hf-027 | +72.17% | 0.777s | 0.216s | 43 | — |
| hf-028 | +72.48% | 0.752s | 0.207s | 41 | — |
| hf-029 | +73.64% | 0.680s | 0.179s | 41 | — |
| hf-030 | +67.93% | 0.533s | 0.171s | 34 | — |
| hf-031 | +56.47% | 0.533s | 0.232s | 42 | — |
| 16 |
| 16.87ms |
| — |
| hf-003 | 76.60ms | 0.279s | 50 | 70.26ms | — |
| hf-004 | 95.60ms | 0.175s | 23 | 16.42ms | — |
| hf-005 | 74.24ms | 0.187s | 43 | 17.01ms | — |
| hf-006 | 112.50ms | 0.285s | 48 | 73.66ms | — |
| hf-007 | 97.08ms | 0.292s | 43 | 72.41ms | — |
| hf-008 | 91.82ms | 0.208s | 43 | 18.13ms | — |
| hf-009 | 86.10ms | 0.199s | 43 | 16.93ms | — |
| hf-010 | 82.37ms | 0.195s | 43 | 16.60ms | — |
| hf-011 | 65.23ms | 0.194s | 43 | 16.42ms | — |
| hf-012 | 73.59ms | 0.187s | 42 | 17.30ms | — |
| hf-013 | 83.24ms | 0.197s | 40 | 17.51ms | — |
| hf-014 | 69.64ms | 0.117s | 18 | 15.96ms | — |
| hf-015 | 99.15ms | 0.336s | 49 | 114.61ms | — |
| hf-016 | 135.74ms | 0.293s | 41 | 70.24ms | — |
| hf-017 | 93.12ms | 0.352s | 67 | 107.69ms | — |
| hf-018 | 122.28ms | 0.186s | 16 | 17.06ms | — |
| hf-019 | 70.88ms | 0.186s | 43 | 18.41ms | — |
| hf-020 | 70.76ms | 0.184s | 43 | 16.61ms | — |
| hf-021 | 61.28ms | 0.127s | 26 | 17.11ms | — |
| hf-022 | 70.32ms | 0.300s | 67 | 73.27ms | — |
| hf-023 | 100.28ms | 0.150s | 18 | 16.42ms | — |
| hf-024 | 69.84ms | 0.158s | 46 | 16.24ms | — |
| hf-025 | 71.39ms | 0.169s | 42 | 16.83ms | — |
| hf-026 | 71.21ms | 0.184s | 43 | 16.86ms | — |
| hf-027 | 71.48ms | 0.288s | 49 | 71.41ms | — |
| hf-028 | 95.45ms | 0.207s | 41 | 16.47ms | — |
| hf-029 | 82.31ms | 0.179s | 41 | 16.54ms | — |
| hf-030 | 72.72ms | 0.291s | 58 | 72.34ms | — |
| hf-031 | 104.50ms | 0.232s | 42 | 16.43ms | — |
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