Cognition launches SWE-2
Coding model post-trained from Kimi K3 with several effort levels, in Devin Desktop and CLI.
It pushes the Pareto frontier of capability and cost, achieving 50.0% on FrontierCode 1.1 Main1, within one point of Fable 5.1 while being 64% cheaper.
Official source
Introducing SWE-2: Pushing the Pareto Frontiercognition.com · published