OpenAI has rolled out a major upgrade to its AI coding agent, Codex, now powered by the latest GPT-5 model. CEO Sam Altman announced the launch on X, highlighting how this advanced tool is designed to make programming faster, smarter, and more efficient.
Smarter and Flexible Performance
The new GPT-5 Codex brings flexibility never seen before. Unlike older versions, it can decide how much time to spend on a task—finishing simple jobs quickly while dedicating up to 7 hours for complex coding problems. This makes it more reliable for projects of all sizes.
Outperforming Previous Versions
OpenAI reports that GPT-5 Codex performs better than the standard GPT-5 in coding benchmarks. It outshines on tests like SWE-bench Verified, which evaluates how well AI can solve real-world coding challenges, and also excels at code refactoring tasks from large repositories.
More Efficient and Developer-Friendly
According to OpenAI’s blog, GPT-5 Codex uses 93.7% fewer tokens compared to GPT-5, making it more cost-efficient. At the same time, it can spend twice as long on reasoning, testing, editing, and coding iterations, delivering higher-quality results.
Smarter Code Review
One of the standout improvements is its ability to review code and detect errors more accurately. GPT-5 Codex provides high-impact comments, helping developers fix issues quickly and write cleaner, more reliable code.
Competitive AI Coding Market
The market for AI coding tools is becoming increasingly competitive, with rivals like Claude Code, Cursor, and Replit gaining traction. With this upgrade, OpenAI aims to maintain its edge by offering more advanced features and broader accessibility.
Availability
The updated GPT-5 Codex is being rolled out in Codex across ChatGPT Plus, Pro, Business, Edu, and Enterprise plans. OpenAI has also confirmed that it will be available soon via API, opening the door for developers and businesses to integrate it directly into their workflows.