OpenAI Upgrades Codex to GPT-5.5 With Expanded Browser and Computer Use

OpenAIOpenAI

· Updated

OpenAI updated its Codex agentic coding platform with GPT-5.5, enabling the agent to autonomously navigate web browsers and control desktop applications. This shift allows developers to automate end-to-end workflows like UI testing and asset generation directly within their engineering environment.

OpenAI upgraded its Codex agentic coding (autonomous software development and iteration) platform to GPT-5.5, which follows the release of frontier models with self-correction capabilities. The agent can now interact with web applications, execute test flows, and capture screenshots. This update also integrates gpt-image-2 for generating UI mockups within the same workflow.
Core model
GPT-5.5
Image generation model
gpt-image-2
Team credit offer
Up to $500
Supported surfaces
IDE, Terminal, Desktop App
New automation features
Worktrees, Skills, Automations
App availability
Waitlist

This release marks a transition from code generation to autonomous engineering execution. By giving the agent direct control over the browser and local files, it can handle complex tasks that previously required manual intervention. It follows a broader industry shift toward computer use where agents operate graphical interfaces.

You can now use the Codex app to run parallel agent workflows using Worktrees, which builds on the multi-agent architecture introduced earlier this year. The platform is available via the IDE, terminal, or a desktop app, which has a waitlist. Teams starting now can earn up to $500 in credits.

View the full update on chatgpt.com
OpenAI Developers
OpenAI Developers
@OpenAIDevs
X

With GPT-5.5, Codex now gets more of the job done across the browser, files, docs, and your computer. We've expanded browser use so Codex can interact with web apps, and test flows, click through pages, capture screenshots, and iterate on what it sees until it completes the task.

171retweets2.2klikes
View on X

Still wondering? A few quick answers below.

Codex is an AI coding agent powered by ChatGPT that helps developers build and ship software. It operates across multiple surfaces, including a desktop app, an IDE extension, and a command-line interface. It is designed to handle end-to-end engineering tasks like feature building, complex refactors, and migrations rather than just providing simple code completions.

With the GPT-5.5 update, Codex can now autonomously interact with web applications and the general computer interface. It can navigate browsers to test flows, click through pages, and capture screenshots to iterate on tasks. This allows the agent to work across files, documents, and web apps to complete complex engineering jobs without manual direction.

Codex now integrates gpt-image-2, allowing users to generate and edit visuals directly while building applications or mockups. Developers can use a specific skill to create assets like landing page hero images or UI components. The agent generates the visual, shows the result, and can provide variations to ensure the design aligns with the project requirements.

Skills are reusable capabilities that help Codex understand a team's specific standards for tasks like prototyping or documentation. Automations allow Codex to work unprompted in the background on routine engineering work. This includes monitoring CI/CD pipelines, triaging issues, and checking for software crashes, allowing developers to focus on high-level system design and high-leverage decisions.

You can use Codex through an IDE extension, the terminal via a command-line interface, or the dedicated Codex app. While the IDE and terminal tools are available for use, the Codex app currently has a waitlist that users can join. Teams starting with Codex may also be eligible for up to $500 in credits to support their workflows.

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

OpenAI

See all AI news & updates from OpenAI →

Share this update