OpenAI has released an upgraded version of Codex powered by the new GPT-5 model, enhancing its capabilities in code generation and understanding. The improvements aim to make the AI more efficient and accurate in assisting developers with programming tasks. This upgrade reflects OpenAI's ongoing commitment to advancing AI technology for practical applications in software development.

OpenAI has launched Codex, a coding tool designed for developers to delegate simple programming tasks to an AI agent that generates production-ready code. Accessible via the ChatGPT web app, Codex operates within a user’s development environment and can be customized through an “AGENTS.md” file for improved effectiveness. Built on a fine-tuned model, Codex leverages reinforcement learning to enhance its coding capabilities.

Codex is a cloud-based software engineering agent designed to assist developers by performing tasks such as writing features, fixing bugs, and generating pull requests in parallel. Available to various user tiers, it prioritizes security and transparency, ensuring users can validate outputs while mitigating risks associated with AI misuse. Early adopters like Cisco and Temporal have successfully integrated Codex into their workflows to enhance productivity and streamline development processes.