$ replaybook

Model results

Pareto 26.10

unbiased/pareto-26.10

Each row is an independent scenario cohort. Releases, providers and reasoning settings remain separate; these results are not pooled into a single model score.

Repairs, timing and cost

ScenarioReleaseProvider / reasoningRepairsMedian timeKnown costEvidence
Reservation write contention v1frontier 20261029.0.0OpenRouter / Default 3/30 unavailable 182s $0.26873/3 trials report cost Inspect attempts →
Notification outbox rollout v1frontier 20261029.0.0OpenRouter / Default 3/30 unavailable 116s $0.22273/3 trials report cost Inspect attempts →
Checkout dependency rollout v1frontier 20261029.0.0OpenRouter / Default 3/30 unavailable 61s $0.13953/3 trials report cost Inspect attempts →

Compare benchmark configurations → · All models