Leaderboard

SWE-Atlas Test Writing

10 models scored · metric: score. Prices are the cheapest listed across all serving providers. Cost/run and Pts/$ are priced for the long context token profile — a whole codebase or filing per request — the case tiered pricing exists for.

A benchmark is priced for the workload it exercises, not a fixed input:output blend: an agent loop and a single hard question bill very differently on the same rate card. Profile: 250k in · 2k out · 30% cache hit · 256k context.

Scores are facts as publicly reported by labs (see report links below and on each model page) and aggregated via models.dev. SWE-Atlas Test Writing is maintained by its own project — Model Pulse is not affiliated with or endorsed by it.