OpenAI Releases GPT-5-Codex Enhancing Autonomous Coding Capabilities

OpenAI has released GPT-5-Codex, an upgraded version of its AI-powered coding assistant optimized specifically for complex software engineering tasks and agentic workflows. This new model can autonomously work on coding projects for extended periods—up to seven hours—iterating and fixing issues without human intervention, and it outperforms the default GPT-5 in benchmarks for code refactoring and agentic coding abilities. GPT-5-Codex has been trained to conduct comprehensive code reviews, catching critical flaws and submitting higher-impact, more accurate review comments, which helps address common challenges like cleaning up poorly written AI-generated code. The update includes new features such as a rebuilt command-line interface focused on agentic workflows, IDE integrations for tools like VS Code, and GitHub integrations for code review, enhancing accessibility through terminals, web, smartphones, and various development environments. Access to GPT-5-Codex is available for ChatGPT Plus, Pro, Business, Edu, and Enterprise users, with API access planned for the near future. OpenAI’s enhancements aim to keep Codex competitive amid a crowded AI coding tool market, where rivals like GitHub Copilot and Cursor have gained significant traction.