SWE-CoWork evaluates whether coding agents can complete end-to-end delivery in software organizations through 101 real code repositories, 26 simulated colleagues, and 101 organization-level tasks. The F2P test pass rates for Claude Opus 5.5 and GPT-6 Astra are approximately 63.84% and 62.91%, respectively, while their full-task pass rates are only 6.93% and 7.92%.
Source:explainx.ai · Read original
Leave a Reply