Release
Cognition turned an open model into a coding beast. But it won’t release the weights.
The New Stack · By Amanda Caswell ·
The Brief
- Cognition released SWE-2 Flash, a coding model, on Wednesday.
- SWE-2 Flash costs one-fifth as much as Cognition's larger SWE-2 model.
- Cognition built the model on an openly downloadable base model but is not releasing its finished weights.
Why it matters
The finished SWE-2 Flash weights cannot be downloaded or self-hosted, so developers who want to run this coding model must rely on Cognition's offering. Openness of the base model does not carry over to this version.
Who’s affected
- Developers who want to self-host coding models
- Teams evaluating SWE-2 Flash for cost-sensitive coding workloads
- #coding-models
- #open-weights
- #cognition
- #ai-models
- #model-release
Cognition built its newest coding model on a foundation that anyone can download, yet the company is keeping its own finished version locked away. Released Wednesday, SWE-2 Flash costs one-fifth as much as Cognition’s larger SWE-2 model and generates tokens roughly three times…