OpenAI expands Codex with computer use on macOS, image generation, and 90+ new plugins

OpenAI Expands Codex Across macOS, Image Generation, and 90+ New Integrations

OpenAI has substantially broadened Codex's capabilities, adding computer use on macOS, native image generation through GPT image 1.5, and over 90 new plugins. The update positions Codex as a more comprehensive automation and design tool rather than a developer-only product.

Computer use on macOS lets Codex interact with any application by seeing, clicking, and typing with its own cursor. The system runs in the background without taking control of the user's machine, handling tasks like frontend iteration, app testing, and workflows that lack API access. This closes a gap for teams relying on legacy tools or closed interfaces.

Image generation and editing integrates GPT image 1.5 directly into Codex's workflow. Users can generate and iterate on images for frontend designs, mockups, and game assets without leaving the tool. Usage is included in ChatGPT accounts with no separate API fee required.

The expansion also includes multi-terminal SSH support, the ability to run automations within a single thread, and 90+ new plugins to connect external tools. Together, these updates enable Codex to handle more stateful, multi-step tasks while learning from prior actions and remembering user preferences for ongoing automation.

The shift from isolated API calls to continuous, stateful automation on everyday applications signals OpenAI's intent to move Codex beyond engineering workflows into general business operations.