Cognition introduces SWE-2 Flash, a coding model with weights withheld
AICognition introduces SWE-2 Flash, a coding model it calls fully American, available today in Devin Desktop and CLI at one-fifth the price and 3x the speed of SWE-2. Cognition says Terminal-Bench 4.0 scores rose from 0.9% to 20.0% over the base model Inkling-Small, and the company is not publicly releasing the weights because of cyber misuse risk, instead planning a program to share them with trusted industry partners and research labs.
AIWhy it matters.
The post gives specific benchmark gains from a mid-training stage and explains why the weights are withheld for cyber-risk reasons, which is useful context for evaluating open-weight release decisions.
Loading source posts…
Source details
Cognition Blog (Devin, Windsurf) · cognition.comPublished · added here
Original title: Introducing SWE-2 Flash