Cognition launches SWE-2 Flash coding model, keeping weights private
Cognition says its SWE-2 Flash beats all American open-weight models, including ones three times larger, and that its weights will stay private.
Cognition has released SWE-2 Flash, a coding model that the company says outperforms all American open-weight models, including ones three times larger. The weights will stay private. According to Jeff Wang, the model is post-trained on Thinking Machines' Inkling-Small, costs 20% as much as SWE-2 to run, and runs about 3x faster than SWE-2.
Silas Alberti said SWE-2 Flash is Cognition's first model to include a mid-training stage and that it was built in six weeks.
Cognition says the weights will not be public. Instead, it plans a secure weight-sharing program for partners and researchers. It cites the risk of cyber misuse as the reason.
The model is free through November for Devin Pro, Max and Teams users.
What changed
Cognition released SWE-2 Flash, a coding model whose weights will stay private, free through November for Devin Pro, Max and Teams users.
Type: New capability · Applies to: SWE-2 Flash, Devin Pro, Max, Teams
Why it matters
Companies using Devin can get a cheaper, faster coding model now. The closed-weight policy means buyers cannot self-host it.
What you can do
Devin customers can trial SWE-2 Flash during the free November window and compare cost and speed against their current coding model.
Original source
We have no link to the original source for this story yet.
Where we saw it
Latent Space
AI wrote this summary from the sources listed, and nobody reviewed it individually before publication. Check the original source before you rely on it. How it works