Details
- Codex now supports computer use on macOS, allowing the AI agent to see, clicking, and type with its own cursor to interact with any app without exposing APIs, useful for frontend iteration, app testing, and custom workflows.
- Image generation capability integrated via gpt-image-1.5, enabling creation of frontend designs, mockups, and game assets directly within Codex without requiring an API key.
- Automations can now run continuously within the same thread, allowing Codex to retain context and schedule future work or wake up automatically to handle long-term tasks like landing pull requests or following up on ongoing items.
- Support for 90+ plugins added, expanding Codex's ability to integrate with tools for documentation, project management, code review, creative work, and deployments.
- Desktop app updates rolling out immediately across Windows and macOS platforms.
Impact
Codex's expansion into computer use and native image generation positions OpenAI to compete more directly with Anthropic's Claude desktop capabilities and Atlassian's AI-powered developer tools. The addition of persistent automations and plugin ecosystem narrows gaps with specialized AI development platforms, potentially accelerating adoption among teams seeking unified AI-assisted workflows. macOS support completion and native image generation reduce reliance on external services, lowering operational friction for developers.
