Round score
0.6948
+69.48%
Prompts scored
25 / 32
median_e2e_speedup
Below tolerance
7
answered too little to score
Zeroed total
7
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-001 | — | — | — | 25 | candidate output below tolerance |
| hf-000 | +65.15% | 0.722s | 0.252s | 42 | — |
| hf-002 | +73.28% | 0.673s | 0.180s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 98.22ms | 0.252s | 42 | 41.29ms | — |
| hf-001 | 84.51ms | 0.134s | 25 | 17.19ms | — |
| hf-002 | 68.30ms | 0.195s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 42 |
| — |
| hf-003 | +70.22% | 0.613s | 0.183s | 43 | — |
| hf-005 | +73.42% | 0.670s | 0.178s | 42 | — |
| hf-004 | — | — | — | 40 | candidate output below tolerance |
| hf-006 | +72.93% | 0.663s | 0.179s | 42 | — |
| hf-007 | — | — | — | 43 | candidate output below tolerance |
| hf-008 | +68.41% | 0.608s | 0.192s | 43 | — |
| hf-010 | +72.72% | 0.710s | 0.194s | 42 | — |
| hf-009 | — | — | — | 45 | candidate output below tolerance |
| hf-011 | +75.08% | 0.716s | 0.178s | 43 | — |
| hf-012 | — | — | — | 43 | candidate output below tolerance |
| hf-013 | — | — | — | 15 | candidate output below tolerance |
| hf-014 | +73.85% | 0.746s | 0.195s | 42 | — |
| hf-015 | +81.48% | 0.920s | 0.171s | 50 | — |
| hf-016 | +59.32% | 0.725s | 0.295s | 41 | — |
| hf-018 | +66.34% | 0.767s | 0.258s | 45 | — |
| hf-017 | +72.05% | 1.098s | 0.307s | 58 | — |
| hf-019 | +69.39% | 0.700s | 0.214s | 39 | — |
| hf-021 | +75.94% | 0.792s | 0.191s | 51 | — |
| hf-022 | +68.33% | 0.358s | 0.113s | 18 | — |
| hf-020 | +72.97% | 0.769s | 0.208s | 44 | — |
| hf-023 | +71.19% | 0.706s | 0.203s | 43 | — |
| hf-024 | +72.93% | 0.751s | 0.203s | 47 | — |
| hf-025 | +75.26% | 0.745s | 0.184s | 43 | — |
| hf-026 | +65.34% | 1.005s | 0.348s | 61 | — |
| hf-027 | +63.93% | 0.893s | 0.322s | 50 | — |
| hf-028 | — | — | — | 18 | candidate output below tolerance |
| hf-029 | +69.57% | 0.697s | 0.212s | 45 | — |
| hf-031 | +52.68% | 0.381s | 0.180s | 28 | — |
| hf-030 | +73.11% | 0.789s | 0.212s | 60 | — |
| 43 |
| 17.45ms |
| — |
| hf-003 | 70.32ms | 0.183s | 43 | 17.94ms | — |
| hf-004 | 65.35ms | 0.192s | 40 | 17.18ms | — |
| hf-005 | 65.98ms | 0.178s | 42 | 17.65ms | — |
| hf-006 | 83.60ms | 0.179s | 42 | 17.33ms | — |
| hf-007 | 66.96ms | 0.178s | 43 | 17.60ms | — |
| hf-008 | 65.12ms | 0.192s | 47 | 17.86ms | — |
| hf-009 | 66.42ms | 0.209s | 45 | 17.41ms | — |
| hf-010 | 98.58ms | 0.194s | 42 | 17.63ms | — |
| hf-011 | 67.38ms | 0.178s | 43 | 17.58ms | — |
| hf-012 | 66.63ms | 0.163s | 43 | 17.32ms | — |
| hf-013 | 65.67ms | 0.115s | 15 | 17.09ms | — |
| hf-014 | 67.86ms | 0.195s | 43 | 17.09ms | — |
| hf-015 | 58.96ms | 0.171s | 53 | 17.98ms | — |
| hf-016 | 93.79ms | 0.295s | 41 | 47.20ms | — |
| hf-017 | 94.95ms | 0.307s | 58 | 42.23ms | — |
| hf-018 | 73.47ms | 0.258s | 45 | 42.55ms | — |
| hf-019 | 102.35ms | 0.214s | 40 | 17.94ms | — |
| hf-020 | 111.54ms | 0.208s | 44 | 17.28ms | — |
| hf-021 | 78.41ms | 0.191s | 51 | 17.67ms | — |
| hf-022 | 63.83ms | 0.145s | 30 | 17.62ms | — |
| hf-023 | 76.29ms | 0.203s | 43 | 17.49ms | — |
| hf-024 | 76.75ms | 0.203s | 47 | 17.72ms | — |
| hf-025 | 88.10ms | 0.184s | 43 | 17.87ms | — |
| hf-026 | 128.94ms | 0.348s | 61 | 61.86ms | — |
| hf-027 | 114.32ms | 0.322s | 50 | 39.80ms | — |
| hf-028 | 88.66ms | 0.138s | 18 | 16.62ms | — |
| hf-029 | 55.16ms | 0.212s | 45 | 17.11ms | — |
| hf-030 | 85.25ms | 0.212s | 60 | 16.96ms | — |
| hf-031 | 84.61ms | 0.197s | 42 | 17.48ms | — |
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