Cognition 推出 SWE-2，为 Devin 提供更便宜的编码模型
Cognition 表示，基于该公司自己的基准测试，SWE-2 在编码性能上接近 Fable 5.1，其部署成本为后者的 36%。
By RuntimeWire Staff · Published
Primary source: Cognition
为何重要
SWE-2 测试 Cognition 是否能将训练后优化和更严格的代理行为转化为 Devin 用户的更低成本。该公司声称节省 64% 的成本基于其自身的基准测试，因此独立评估将是下一项有意义的检验。
Scott Wu's Cognition, which operates Devin, released SWE-2 on September 10th. The model is available in Devin Desktop and its 命令行界面, with a rollout underway across Devin Web and Fusion. Cognition says SWE-2 approaches the coding performance of Fable 5.1 and GPT-6 Astra at substantially lower rollout costs.…
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.