Codex for Almost Everything
Key point
OpenAI has unveiled a major update to the Codex app that automates developer workflows by integrating computer control and image generation capabilities.
Details
OpenAI's Codex app has received a major update for macOS and Windows, evolving beyond simple code writing into an integrated workspace that supports the entire software development process.
Key updates:
- Computer control and parallel tasks: Added the ability to click, type, and navigate directly on the user's computer. It can perform testing or repetitive tasks even in apps without an API, and multiple agents can work in parallel simultaneously.
- Image generation and visual workflows: Integrated the gpt-image-1.5 model, allowing product concepts, frontend designs, game visuals, and more to be created within the same workflow as code writing.
- Plugins and extensibility: Supports over 90 new plugins, including Atlassian Rovo, CircleCI, and GitLab Issues, to share context with various tools and perform automated tasks.
- Enhanced developer-focused features: Supports web page annotation via an in-app browser, PR review, multiple terminal and file views, and remote devbox connection via SSH (alpha).
- Long-running tasks and memory: Reuses existing conversation threads and can self-schedule tasks to sustain long-running work spanning days to weeks. It also features a memory preview function that remembers user preferences and past context.
It is currently rolling out sequentially to ChatGPT account users, with personalization and computer control features expected to expand gradually depending on region and account type.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.