OpenAI has launched Codex, an AI coding agent designed to assist software engineers in writing code and fixing bugs.

Codex operates as a cloud-based virtual coworker, enabling engineers to run multiple tasks in parallel. 

Powered by the o3 AI reasoning model, Codex is optimized for software engineering tasks. 

The AI agent is available to ChatGPT Pro, Enterprise, and Team subscribers. 

Codex can autonomously generate code from natural language prompts, fix bugs, and run tests. 

It operates within a controlled sandboxed environment to maintain security. 

OpenAI engineers are already using Codex internally to handle repetitive and clearly defined assignments. 

Codex's launch positions OpenAI to compete directly with AI coding tools from Google and Anthropic.