A new open-source AI model named Devin has shown promising results in software engineering benchmarks. Devin achieved 12.29% accuracy on 100% of the SWE Bench test set, compared to 13.84% on 25% of the set. The model uses GPT4 and is expected to improve with GPT5. The project is led by a team including John and has garnered attention for its potential in generalization.
Open source with results close to Devin #ai #coding https://t.co/eNxX1dk8FO
Exciting open source Devin from Princeton https://t.co/2eiigNO0Pn
Open source Devin, with very impressive numbers. From the thread: 'letting SWE-agent only view 100 lines at a time was better than letting it view 200 or 300 lines and much better than letting it view the entire file'. https://t.co/mQsplpewWw
An open source Devin getting 12.29% on 100% of the SWE Bench test set vs Devin's 13.84% on 25% of the test set! This uses GPT4, so imagine what GPT5 could do! Although this works on Github repo issues, I'm assuming the next step is a full generalization! Great work John and team! https://t.co/BQW4mp7koN
OPEN SOURCE AND DECENTRALIZE AI !!!!!!!!!!!!! https://t.co/p3Gi9C2188
decentralized ai is a joke! https://t.co/m7nbV9Vf6L https://t.co/uf8VZyXxUc https://t.co/0kOiH4BfWV