OpenAI announced the launch of Codex, its most capable AI coding agent yet, on Friday. Powered by the codex-1 model, which is optimized for software engineering tasks, Codex promises to generate cleaner, more efficient code while adhering more precisely to instructions. The AI will iteratively run tests on its code to ensure it passes successfully.
How Codex Works
Codex runs in a sandboxed, virtual computer in the cloud and integrates seamlessly with GitHub to access your code repositories. It can handle multiple software engineering tasks simultaneously, such as writing simple features, fixing bugs, answering questions about your codebase, and running tests. Tasks are expected to be completed in as little as one to thirty minutes.
Codex Rollout and Access
Starting today, Codex is available to ChatGPT Pro, Enterprise, and Team subscribers. OpenAI has stated that users will initially receive “generous access” to the tool, but rate limits will be implemented in the coming weeks, with users able to purchase additional credits. OpenAI plans to expand Codex access to ChatGPT Plus and Edu users soon.
Rising Popularity of AI Coding Tools
The rise of AI tools for software engineers, or “vibe coders,” has been a notable trend. Industry leaders such as Google and Microsoft have claimed that a significant portion of their code is now being written by AI. In response, OpenAI is looking to capture a piece of the growing market, further evidenced by their recent acquisition of Windsurf, a developer behind another popular AI coding platform, for $3 billion.
Safety Measures and Future Goals
Codex will operate in a controlled, air-gapped environment with no access to the broader internet, ensuring that it cannot develop malicious software. OpenAI’s approach to safety is a key aspect of its design, with the AI refusing requests to produce harmful code. However, like all AI systems, Codex is not immune to errors, with studies showing that other AI coding models sometimes struggle to reliably debug software.
Codex CLI and Pricing
In addition to Codex, OpenAI has updated its open-source Codex CLI, which runs directly in a terminal. The new CLI version utilizes the o4-mini model, optimized for software engineering tasks. This model is now the default in Codex CLI and will be available through OpenAI’s API at $1.50 per million input tokens and $6 per million output tokens.
OpenAI’s Expanding Product Line
Codex is part of OpenAI’s broader strategy to enhance ChatGPT with more products beyond the popular chatbot. In the past year, OpenAI has introduced AI-powered video platform Sora, the research agent Deep Research, and web browsing agent Operator as benefits for subscribers. These offerings, along with Codex, are likely to attract more users and encourage existing subscribers to pay for additional features.
