Highlights §
- With our advances in long-term reasoning and planning, Devin can plan and execute complex engineering tasks requiring thousands of decisions. Devin can recall relevant context at every step, learn over time, and fix mistakes.
We’ve also equipped Devin with common developer tools including the shell, code editor, and browser within a sandboxed compute environment—everything a human would need to do their work.
Finally, we’ve given Devin the ability to actively collaborate with the user. Devin reports on its progress in real time, accepts feedback, and works together with you through design choices as needed. (View Highlight)
- With our advances in long-term reasoning and planning, Devin can plan and execute complex engineering tasks requiring thousands of decisions. Devin can recall relevant context at every step, learn over time, and fix mistakes.
We’ve also equipped Devin with common developer tools including the shell, code editor, and browser within a sandboxed compute environment—everything a human would need to do their work.
Finally, we’ve given Devin the ability to actively collaborate with the user. Devin reports on its progress in real time, accepts feedback, and works together with you through design choices as needed. (View Highlight)
- **Devin can learn how to use unfamiliar technologies.
**After reading a blog post, Devin runs ControlNet on Modal to produce images with concealed messages for Sara. (View Highlight)
- **Devin can build and deploy apps end to end.
**Devin makes an interactive website which simulates the Game of Life! It incrementally adds features requested by the user and then deploys the app to Netlify. (View Highlight)
- Devin’s Performance
We evaluated Devin on SWE-bench, a challenging benchmark that asks agents to resolve real-world GitHub issues found in open source projects like Django and scikit-learn.
Devin correctly resolves 13.86%* of the issues end-to-end, far exceeding the previous state-of-the-art of 1.96%. Even when given the exact files to edit, the best previous models can only resolve 4.80% of issues. (View Highlight)