OpenAI has officially begun rolling out its enhanced GPT-5 Codex model across all Codex platforms, including Terminal, IDE extensions, and the web-based Codex interface at chatgpt.com/codex. This significant update aims to strengthen OpenAI’s position in the competitive AI coding market, where it currently trails behind Claude Code.
## What is Codex?
Codex is OpenAI’s AI-powered coding assistant that automates programming tasks. The platform allows users to delegate complex coding challenges to the AI, which then executes code autonomously. Notably, even users without programming experience can leverage Codex to create applications and web apps through intuitive “vibe coding” – a more conversational approach to software development.
## Impressive Performance Gains
The new GPT-5 Codex model demonstrates substantial improvements in real-world coding scenarios. According to OpenAI’s official blog post, the upgraded model achieves a 74.5% success rate on the SWE-bench Verified benchmark, a significant leap in coding accuracy.
The improvements are particularly notable in code refactoring tasks, where performance jumped from 33.9% with the standard GPT-5 model to an impressive 51.3% with GPT-5 Codex – representing a 51% improvement in refactoring capabilities.
## Gradual Rollout in Progress
While the announcement has generated excitement among developers, the rollout is happening gradually. Some users, including ChatGPT Plus subscribers paying $20 monthly, report not yet having access to the new features. OpenAI has confirmed that the GPT-5 Codex model will be fully available to all users within the coming days.
## Market Competition
This release represents OpenAI’s strategic move to compete more effectively with Claude Code, which currently leads the AI coding assistant market. The enhanced performance metrics suggest that GPT-5 Codex could significantly narrow this gap and provide developers with a more capable alternative for automated programming tasks.
The upgrade signals OpenAI’s commitment to advancing AI-assisted development tools, potentially making programming more accessible to a broader audience while improving productivity for experienced developers.
