Cognition's SWE-2

SWE-2 is Cognition's coding model, post-trained from Kimi K3 with reinforcement learning that optimizes for cost and capability simultaneously. According to Cognition, it scores 50.0% on FrontierCode 1.1 Main, within a point of Fable 5.1 at 64% lower cost, and near GPT-6 Astra at a quarter of the cost. Compared to SWE-1.7, it takes 58% fewer turns, costs 81% less, and scores higher. It is available now in Devin Desktop and CLI.