Round score
0.7871
+78.71%
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 | +72.60% | 0.592s | 0.162s | 42 | — |
| hf-001 | +74.75% | 0.506s | 0.128s | 28 | — |
| hf-003 | +78.39% | 0.751s | 0.162s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 57.81ms | 0.162s | 42 | 18.04ms | — |
| hf-001 | 50.45ms | 0.168s | 43 | 19.04ms | — |
| hf-002 | 51.94ms | 0.199s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 41 |
| — |
| hf-002 | +82.02% | 1.108s | 0.199s | 63 | — |
| hf-005 | +81.15% | 0.579s | 0.109s | 31 | — |
| hf-004 | +83.53% | 0.923s | 0.152s | 49 | — |
| hf-006 | +81.80% | 1.009s | 0.184s | 61 | — |
| hf-008 | +78.96% | 0.732s | 0.154s | 42 | — |
| hf-007 | +79.89% | 0.544s | 0.109s | 31 | — |
| hf-009 | +82.54% | 0.794s | 0.139s | 48 | — |
| hf-010 | +79.28% | 0.951s | 0.197s | 53 | — |
| hf-011 | +79.22% | 0.781s | 0.162s | 43 | — |
| hf-012 | +76.78% | 0.720s | 0.167s | 42 | — |
| hf-014 | +80.92% | 0.730s | 0.139s | 42 | — |
| hf-016 | +71.76% | 0.341s | 0.096s | 16 | — |
| hf-018 | +68.18% | 0.401s | 0.128s | 26 | — |
| hf-017 | +80.55% | 0.724s | 0.141s | 41 | — |
| hf-015 | — | — | — | 55 | candidate output below tolerance |
| hf-019 | +78.46% | 0.716s | 0.154s | 41 | — |
| hf-020 | +78.08% | 0.772s | 0.169s | 46 | — |
| hf-013 | — | — | — | 43 | candidate output below tolerance |
| hf-021 | +78.16% | 0.703s | 0.154s | 42 | — |
| hf-022 | — | — | — | 21 | candidate output below tolerance |
| hf-023 | +81.21% | 0.732s | 0.138s | 41 | — |
| hf-024 | +70.85% | 0.676s | 0.197s | 42 | — |
| hf-025 | +76.99% | 0.762s | 0.175s | 45 | — |
| hf-026 | +80.12% | 0.765s | 0.152s | 47 | — |
| hf-028 | +80.98% | 0.763s | 0.145s | 43 | — |
| hf-027 | +82.00% | 1.006s | 0.181s | 62 | — |
| hf-029 | +80.88% | 0.731s | 0.140s | 46 | — |
| hf-030 | +78.40% | 0.710s | 0.153s | 51 | — |
| hf-031 | +71.57% | 0.636s | 0.181s | 51 | — |
| 63 |
| 15.77ms |
| — |
| hf-003 | 60.10ms | 0.162s | 41 | 14.87ms | — |
| hf-004 | 49.95ms | 0.152s | 49 | 14.82ms | — |
| hf-005 | 50.83ms | 0.109s | 31 | 14.81ms | — |
| hf-006 | 52.28ms | 0.184s | 61 | 15.22ms | — |
| hf-007 | 52.41ms | 0.109s | 31 | 14.89ms | — |
| hf-008 | 51.38ms | 0.154s | 42 | 18.69ms | — |
| hf-009 | 51.30ms | 0.139s | 48 | 15.03ms | — |
| hf-010 | 51.52ms | 0.197s | 53 | 15.16ms | — |
| hf-011 | 61.34ms | 0.162s | 43 | 14.71ms | — |
| hf-012 | 51.63ms | 0.167s | 42 | 15.08ms | — |
| hf-013 | 50.88ms | 0.167s | 43 | 14.97ms | — |
| hf-014 | 51.61ms | 0.139s | 42 | 15.08ms | — |
| hf-015 | 50.22ms | 0.167s | 55 | 14.88ms | — |
| hf-016 | 52.73ms | 0.171s | 44 | 15.80ms | — |
| hf-017 | 53.74ms | 0.141s | 41 | 15.08ms | — |
| hf-018 | 55.24ms | 0.128s | 28 | 14.86ms | — |
| hf-019 | 52.80ms | 0.154s | 41 | 15.14ms | — |
| hf-020 | 52.99ms | 0.169s | 46 | 15.29ms | — |
| hf-021 | 51.87ms | 0.154s | 42 | 14.94ms | — |
| hf-022 | 51.93ms | 0.124s | 21 | 14.84ms | — |
| hf-023 | 50.50ms | 0.138s | 41 | 15.11ms | — |
| hf-024 | 51.41ms | 0.197s | 42 | 14.86ms | — |
| hf-025 | 59.34ms | 0.175s | 45 | 14.90ms | — |
| hf-026 | 50.43ms | 0.152s | 47 | 14.91ms | — |
| hf-027 | 50.21ms | 0.196s | 64 | 14.67ms | — |
| hf-028 | 56.27ms | 0.145s | 43 | 15.58ms | — |
| hf-029 | 51.01ms | 0.140s | 46 | 15.59ms | — |
| hf-030 | 51.69ms | 0.153s | 51 | 15.05ms | — |
| hf-031 | 50.30ms | 0.181s | 51 | 15.35ms | — |
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