April 17 news.OpenAI A lightweight terminal runtime code was released todayAgent —— Codex CLIThe tool is now available at GitHub whollyOpen Source.
The Codex CLI works directly on the user's computer.Designed to maximize the inference power of models such as o3 and o4-miniand will soon support additional API models such as GPT-4.1.
Users have access to multimodal reasoning via the command line, e.g. by passing screenshots or low-fidelity sketches to the model, combined with access to native code.
OpenAI considers the Codex CLI to be a minimalist interface that connects its models to users and their computers.The Codex CLI is designed for developers who already live in a terminalThey want ChatGPT-level reasoning capabilities, and the power to actually run code, manipulate files, and iterate -- all under version control. In short, it's a chat-driven development tool that understands and executes repositories.
- Zero Configuration - Import the OpenAI API key and you're ready to go!
- Fully automated approvals, also secured by running network disabling and directory sandboxing
- Multimodal - Input screenshots or charts to enable reasoning
The Codex CLI can be used on macOS 12+, Ubuntu 20.04+/Debian 10+, and the WSL2 subsystem of Windows 11 and requires a minimum of 4GB of RAM (8GB recommended).
1AI Attached is a link to the Codex CLI open source:
https://github.com/openai/codex