Round score
0.0069
+0.69%
Prompts scored
30 / 32
median_e2e_speedup
Below tolerance
2
answered too little to score
Zeroed total
2
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 | +0.46% | 0.489s | 0.486s | 28 | — |
| hf-001 | +0.70% | 0.871s | 0.865s | 52 | — |
| hf-002 | +0.78% | 0.788s | 0.782s |
32 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 95.96ms | 0.486s | 28 | 92.02ms | — |
| hf-001 | 103.00ms | 0.865s | 52 | 94.17ms | — |
| hf-002 | 64.88ms | 0.782s |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
| 51 |
| — |
| hf-003 | -0.29% | 0.740s | 0.743s | 43 | — |
| hf-004 | +1.37% | 0.740s | 0.730s | 43 | — |
| hf-005 | +2.46% | 0.791s | 0.772s | 47 | — |
| hf-006 | +0.31% | 0.706s | 0.703s | 40 | — |
| hf-007 | +0.50% | 0.738s | 0.734s | 43 | — |
| hf-009 | +0.11% | 0.805s | 0.804s | 52 | — |
| hf-008 | +0.69% | 1.123s | 1.115s | 68 | — |
| hf-010 | +1.02% | 0.755s | 0.747s | 44 | — |
| hf-011 | +1.10% | 0.715s | 0.707s | 40 | — |
| hf-012 | +0.60% | 0.935s | 0.930s | 50 | — |
| hf-014 | +0.33% | 0.758s | 0.755s | 41 | — |
| hf-013 | +1.04% | 1.009s | 0.999s | 57 | — |
| hf-015 | +1.22% | 0.751s | 0.742s | 40 | — |
| hf-016 | +1.63% | 0.783s | 0.770s | 43 | — |
| hf-017 | +1.00% | 0.803s | 0.795s | 45 | — |
| hf-018 | +0.30% | 0.779s | 0.777s | 43 | — |
| hf-020 | -0.58% | 0.728s | 0.732s | 42 | — |
| hf-019 | +0.58% | 1.013s | 1.008s | 61 | — |
| hf-021 | +1.05% | 0.704s | 0.697s | 43 | — |
| hf-022 | +1.07% | 0.599s | 0.592s | 40 | — |
| hf-023 | — | — | — | 36 | candidate output below tolerance |
| hf-024 | +1.22% | 0.728s | 0.720s | 42 | — |
| hf-025 | — | — | — | 42 | candidate output below tolerance |
| hf-026 | +0.77% | 0.780s | 0.774s | 43 | — |
| hf-027 | +0.73% | 0.799s | 0.793s | 48 | — |
| hf-029 | -0.05% | 0.645s | 0.645s | 41 | — |
| hf-028 | +0.36% | 0.848s | 0.845s | 53 | — |
| hf-030 | +1.24% | 0.634s | 0.626s | 47 | — |
| hf-031 | -0.01% | 0.537s | 0.537s | 42 | — |
| 51 |
| 94.15ms |
| — |
| hf-003 | 110.26ms | 0.743s | 43 | 93.65ms | — |
| hf-004 | 100.69ms | 0.730s | 43 | 93.63ms | — |
| hf-005 | 64.69ms | 0.783s | 48 | 93.65ms | — |
| hf-006 | 110.16ms | 0.703s | 40 | 90.88ms | — |
| hf-007 | 107.47ms | 0.734s | 43 | 93.65ms | — |
| hf-008 | 105.36ms | 1.115s | 68 | 93.68ms | — |
| hf-009 | 73.21ms | 0.804s | 52 | 94.16ms | — |
| hf-010 | 102.83ms | 0.747s | 44 | 93.63ms | — |
| hf-011 | 65.98ms | 0.707s | 40 | 93.61ms | — |
| hf-012 | 106.04ms | 0.930s | 50 | 94.05ms | — |
| hf-013 | 104.12ms | 0.999s | 57 | 94.13ms | — |
| hf-014 | 107.45ms | 0.755s | 41 | 94.27ms | — |
| hf-015 | 105.67ms | 0.742s | 40 | 93.91ms | — |
| hf-016 | 102.78ms | 0.770s | 43 | 93.53ms | — |
| hf-017 | 101.88ms | 0.795s | 45 | 93.57ms | — |
| hf-018 | 104.71ms | 0.777s | 43 | 93.87ms | — |
| hf-019 | 106.47ms | 1.008s | 61 | 93.70ms | — |
| hf-020 | 110.10ms | 0.732s | 42 | 93.61ms | — |
| hf-021 | 102.21ms | 0.697s | 43 | 93.68ms | — |
| hf-022 | 65.40ms | 0.592s | 40 | 93.68ms | — |
| hf-023 | 111.34ms | 0.590s | 36 | 89.41ms | — |
| hf-024 | 64.64ms | 0.720s | 42 | 95.60ms | — |
| hf-025 | 105.04ms | 0.764s | 42 | 95.56ms | — |
| hf-026 | 105.97ms | 0.774s | 43 | 93.72ms | — |
| hf-027 | 107.01ms | 0.793s | 48 | 93.75ms | — |
| hf-028 | 104.69ms | 0.845s | 53 | 93.58ms | — |
| hf-029 | 107.23ms | 0.645s | 41 | 89.27ms | — |
| hf-030 | 64.68ms | 0.626s | 47 | 89.34ms | — |
| hf-031 | 104.34ms | 0.537s | 42 | 12.34ms | — |
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