In a world where time matters, large language models (LLMs) have proven themselves invaluable in helping users accomplish tasks more efficiently.
Since OpenAI introduced ChatGPT, it has sparked an arms race among tech companies of all sizes, each striving to develop the most advanced AI tools. And as these models evolved, one of their most impressive breakthroughs has been their ability to tackle complex tasks in mathematics and programming.
And here, in the latest entrant in this race is OpenAI's Codex, a cutting-edge AI coding agent integrated into ChatGPT.
Codex is designed to help users in software development process by automating various coding tasks, thereby enhancing productivity and efficiency.
Whether users are seasoned developers or novice, OpenAI wants to make Codex an invaluable companion in users' coding journey.
OpenAI introduced Codex in a livestream:
Introducing a research preview of Codex in ChatGPThttps://t.co/EOvjGJrHbj
— OpenAI (@OpenAI) May 16, 2025
According to OpenAI in the announcement:
This parallel execution model allows developers to offload repetitive or well-defined tasks to Codex, enabling them to focus on higher-level design and problem-solving.
For instance, while Codex handles tasks like refactoring code, writing tests, or generating documentation, developers can continue working on other aspects of the project. Codex's ability to work autonomously and in parallel transforms it into a virtual coworker, streamlining the development process and boosting productivity.
As for the brain behind this Codex, is 'codex-1,' a specialized version of OpenAI's o3 reasoning model, optimized for software engineering.
The model was trained with reinforcement learning on real-world coding challenges across diverse environments. Because the model is optimized for coding, OpenAI said that codex-1 can produce "cleaner" code than the o3, follows instructions more accurately, and can run tests until all pass successfully.
OpenAI said that the AI coding agent can take anywhere from one to 30 minutes to write simple features, fix bugs, answer questions about users' codebase, and run tests, among other tasks.
Codex independently navigates your codebase, implements and tests code changes, and proposes pull requests for you to review.
It’s powered by codex-1, a version of OpenAI o3 optimized for software engineering.https://t.co/axMnewGMtT— OpenAI (@OpenAI) May 16, 2025
Codex is initially rolling out to subscribers to ChatGPT Pro, Enterprise, and Team.
OpenAI said that users shall receive "generous access" to Codex to start, but in later, the company plans to implement rate limits for the tool.
In other words, users will then have the option to purchase additional credits to use Codex.
Users with access to Codex can find the tool in ChatGPT’s sidebar, and assign the agent new coding tasks by typing a prompt and clicking the “Code” button. Users can also ask questions about their codebase and click the “Ask” button. Below the prompting bar, users can see other tasks they’ve assigned Codex to do, and monitor their progress.
AI tools for software engineers, often called "vibe coders," have seen a rapid surge in popularity recently.
Tech giants like Google and Microsoft report that around 30% of their code is now generated by AI. Among others, Anthropic with Claude Code and Google’s Gemini Code Assist, have also introduced advanced AI coding assistants, fueling this fast-growing market.
With the rise of AI coding platforms, AI is now making its move.
Besides announcing Codex, OpenAI is also enhancing 'Codex CLI,' an open-source coding agent, with its latest o4-mini model optimized for software engineering, available through its API with transparent pricing.
This launch is part of OpenAI’s broader strategy to expand ChatGPT’s capabilities beyond chat, alongside products like its AI video platform Sora, research agent Deep Research, and browsing agent Operator. These innovations aim to attract new subscribers and encourage existing users to invest more for enhanced features and higher usage limits.