OpenAI has launched Codex, a software engineering agent that can handle multiple tasks in parallel within isolated sandbox environments. These tasks include writing code, fixing bugs, answering codebase questions, and making pull requests. Codex runs on the cloud and is powered by the optimized o3 model, known as codex-1. It has shown strong performance in SWE-Bench and OpenAI’s internal benchmark tests, generating code in a style similar to humans.🧵1/4

openai.com/index/introducing-c…
#AI Threads