Rivalry Heats Up: OpenAI Overhauls Codex to Challenge Anthropic’s Claude Code in the AI Coding Race
An intense, high-stakes competition is currently unfolding between OpenAI and Anthropic, as the two leading AI developers race to launch the most accessible, powerful AI coding tools on the market. To date, Anthropic has pulled ahead: as TechCrunch reported last week, Claude Code has become the go-to preferred tool for many businesses. But OpenAI has no plans to concede the race.
This week, OpenAI unveiled a major overhaul of Codex, its in-house automated coding tool, rolling out a suite of new updates built to dramatically expand the platform’s capabilities.
Announced this past Thursday, the broad slate of new features is headlined by a standout breakthrough: Codex can now run fully in the background on a user’s computer, opening any desktop app and executing actions via an automated cursor that clicks and types on the user’s behalf.
Per an official blog post from OpenAI, this new architecture lets Codex deploy multiple independent AI agents that all work on a user’s Mac in parallel, without disrupting the user’s own work in other applications. Put simply, Codex’s background operation means users can continue using their device as normal while AI agents handle assigned tasks. OpenAI positions the updated tool as a collaborative coding partner that tackles auxiliary work while users focus on high-priority core projects, listing common use cases such as iterating on frontend changes, testing full applications, and working within tools that do not offer public API access.
Taken as a whole, this agent-focused upgrade and other new additions make clear OpenAI’s core goal: not only to position Codex as a competitive rival to top coding assistants, but also to build a versatile, multifaceted tool that can integrate seamlessly into a wide range of corporate workflows.
Observers of the fast-growing AI coding space will quickly notice that many of the new capabilities OpenAI added to Codex mirror features Anthropic previously launched for its own Claude Code offering. Just last month, Anthropic announced that Claude and its Cowork agent can remotely control a user’s Mac and desktop on their behalf, even when the user is away from their keyboard.
Beyond agentic automation tools, the updated Codex now includes a built-in in-app browser, which lets users issue commands that the AI executes directly on specific web applications. OpenAI notes this feature will be particularly useful for frontend and game development, and that the company plans to eventually expand the capability so Codex can “fully command the browser beyond web applications on localhost.”
Meet your next investor or portfolio startup at Disrupt
Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410.
Meet your next investor or portfolio startup at Disrupt
Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410.
The Codex revamp includes several additional key updates:
A new in-preview feature called “Memory” lets Codex recall details from previous work sessions and build tailored context around a specific user’s working style.
The tool has also gained native image-generation capabilities, which OpenAI says can be used to create everything from product concepts and presentation slide visuals to design mockups, placeholder images, and other common business materials.
To expand Codex’s ability to complete end-to-end tasks, OpenAI has added 111 plugin integrations with third-party apps including CodeRabbit and GitLab Issues, enabling the tool to handle cross-tool workflows natively.
OpenAI frames these new plugins as giving Codex the ability to handle routine clerical work to organize a user’s work life. For example, if you ask Codex to scan your Slack channels and Google Calendar to pull together a daily to-do list, the company confirms the tool can now complete that task automatically.
OpenAI also launched a new pay-as-you-go pricing option for Codex, available to all ChatGPT enterprise and business customers. The new tier is designed to give users more flexibility when accessing the coding tool’s services.
Once considered the undisputed leader of the global generative AI industry, OpenAI has ramped up its competition with Anthropic in recent months, shifting its core focus to building enterprise-grade AI capabilities and winding down experimental consumer tools like its social video app Sora 2. The company has also navigated a string of high-profile controversies in recent months, including multiple lawsuits over claims that ChatGPT caused harmful mental health impacts for some users.
Rivalry Heats Up: OpenAI Overhauls Codex to Challenge Anthropic’s Claude Code in the AI Coding Race
