ArtificialAnalysis/Coding Agent Index/Cost per Task
- Mean pay-per-token API cost per task (USD); lower is better.
- Averaged across the same benchmark suite as ArtificialAnalysis/Coding Agent Index: SWE-Bench-Pro-Hard-AA, Terminal-Bench v2, and SWE-Atlas-QnA.
- Pricing includes standard input, discounted cached input, cache-write charges, and output where the provider exposes them — not subscription plan pricing or production infrastructure and supervision cost.
- Parent: ArtificialAnalysis/Coding Agent Index