OpenAI Updates Codex with Agentic Desktop Control and Super App Features
OpenAI has released a major update to its Codex coding agent, introducing advanced agentic capabilities that allow the AI to interact directly with desktop applications, particularly on macOS. This update features background computer control, an integrated web browser for visual feedback, image generation via gpt-image-1.5, and enhanced memory functions. By enabling parallel task execution and integrating numerous plugins, OpenAI aims to transform Codex into a comprehensive "super app" for developers. This strategic move intensifies competition with Anthropic’s Claude Code, marking a significant shift toward autonomous, multi-modal AI workflows in software development.
Cross-source coverage
Wire timeline
OpenAI Revamps Codex with Desktop Control to Rival Anthropic
OpenAI has announced a significant update to its automated coding tool, Codex, intensifying its competition with Anthropic in the AI-coding market. The revamped Codex now features agentic capabilities that allow it to operate in the background on user desktops, specifically macOS, enabling it to open applications and perform cursor-based actions without interfering with the user's primary tasks. This functionality positions Codex as a parallel assistant for tasks like frontend iteration and testing. The update also introduces an in-app browser for executing commands within web applications, a new memory feature for recalling past work contexts, and image-generation abilities for creating visual assets. Additionally, OpenAI integrated 111 plug-ins with platforms such as Slack, Google Calendar, and GitLab, allowing Codex to manage clerical workflows and organize daily tasks. These enhancements reflect OpenAI's strategy to transform Codex into a multifaceted tool embedded in corporate workflows, directly countering Anthropic's recent advances with Claude Code, which offers similar remote desktop control features.
TechCrunchOpenAI Updates Codex Desktop App with Cross-Application Agent Capabilities
OpenAI has released a major update to its Codex desktop application for Mac and Windows, transforming it into a comprehensive 'Super App' for developers. Now serving 3 million weekly users, Codex expands beyond code generation to interact directly with other applications on a user's computer. The standout feature, 'Computer Use,' currently exclusive to macOS, allows the AI agent to see, click, and type within any app in the background, enabling multi-agent workflows without interrupting the user. Windows users gain access to app data integration but lack background cursor control at launch. The update also introduces an integrated web browser for frontend previews and connects to OpenAI’s gpt-image-1.5 model for generating visual assets like mockups and game elements. This strategic move positions Codex as a unified workspace for the entire software development lifecycle, distinguishing it from rival Anthropic’s Claude by offering deeper, simultaneous system-level interaction. OpenAI executives stated that Codex was chosen for these advanced capabilities because it is their most powerful agent already embedded in developers' workflows.
VentureBeatOpenAI Updates Codex with Background Computer Control and In-App Browser
OpenAI has released a significant update to its Codex desktop application, introducing advanced capabilities that allow the AI to operate computers in the background without interfering with user activities. This new feature enables Codex to see, click, and type across various applications using its own cursor, facilitating parallel task execution for developers and non-developers alike. The update also includes an in-app web browser designed for visual feedback on web development projects, allowing users to comment directly on specific page elements. Additionally, Codex can now schedule tasks for future execution and leverage the gpt-image-1.5 model for image generation in mockups. The release incorporates 90 new plugins, expanding its utility beyond software development into general knowledge work. OpenAI executives describe these changes as strategic steps toward building a comprehensive 'super app' that integrates agentic tools, the Atlas web browser, and Codex. This evolution aims to create a unified platform for diverse digital contexts, marking a shift from specialized coding assistance to broader autonomous computer interaction and workflow management.
Ars Technica - All contentOpenAI Updates Codex with Agentic Capabilities to Challenge Anthropic
OpenAI has announced a significant update to its agentic coding and development system, Codex, positioning it as a direct competitor to Anthropic’s Claude Code. The enhanced system now features advanced capabilities that allow it to interact more deeply with user environments. Specifically, Codex can now operate desktop applications on a user's computer, generate images, and retain memory from past interactions to improve contextual understanding. This suite of updates marks a strategic move by OpenAI to strengthen its foothold in the AI-assisted development market. By enabling Codex to perform actions directly within desktop apps, OpenAI aims to provide a more seamless and powerful workflow for developers. The announcement, detailed in an official blog post, highlights the company's focus on creating more autonomous and versatile AI agents. This development underscores the intensifying rivalry between major AI firms as they race to deliver superior coding assistants. The integration of memory and multi-modal capabilities like image generation further expands Codex's utility, potentially transforming how software development tasks are automated and executed in professional settings.
The VergeOpenAI Updates Codex with New Features for Upcoming Super App
OpenAI has released a major update to its Codex coding agent, laying the groundwork for a future desktop super app that will integrate ChatGPT, Codex, and the Atlas web browser. While the unified application is not yet available, this update significantly expands Codex capabilities for developers. Key features include enhanced computer use, allowing AI agents to interact with other applications on macOS without system lag, and the introduction of 111 new plugins for better context gathering. The update also adds a built-in browser with a commenting system for direct webpage adjustments, alongside integrated image generation using gpt-image-1.5 for creating design assets. Additionally, OpenAI previewed new memory features that enable Codex to recall past contexts and suggest proactive actions, such as responding to document comments. Currently, the update is rolling out to desktop users logged into ChatGPT, with computer use initially limited to macOS. Availability for users in the EU and UK, particularly for memory features, will follow later. This release signals OpenAI's strategy to build its super app ecosystem openly, starting with developer-focused tools before expanding to a broader audience.
Engadget is a web magazine with obsessive daily coverage of everything new in gadgets and consumer electronics