OpenAI spills technical details about how its AI coding agent works

OpenAI’s Codex coding agent can write code, run tests and fix bugs under human supervision, making it useful for rapid prototyping but requiring oversight before production use. The Allen Institute for AI’s open-source SERA agents, including 32-billion- and 8-billion-parameter models, are designed to adapt to private codebases and assist developers with a range of coding tasks.

本文综合以下全部报道写成，而非依据任何单一来源。