OpenAI Expands Codex Desktop Into Full Productivity Suite With Computer Control and Image Generation

OpenAI has shipped a significant update to its Codex desktop application, expanding its capabilities well beyond software development. The updated Codex can now control the computer directly — clicking, typing, and navigating applications — operate an in-app browser, generate images, maintain automation memory across sessions, and support plugins that extend its capability into external services. Rate limits have been doubled across all paid plans. The update repositions Codex from a coding assistant into a general-purpose AI agent for knowledge work.
Computer Control: What It Means in Practice
Computer control is the capability that changes Codex's category. A coding assistant generates code; an agent with computer control can run that code, observe the output, navigate to the relevant documentation if something breaks, and iterate — all without the user switching context. More broadly, computer control allows Codex to automate workflows that previously required either custom scripting or manual repetition: filling out forms, extracting data from applications that don't have APIs, reorganizing files, or composing and sending communications. This capability puts Codex in direct competition with Anthropic's Claude computer use and Adept's AI.
The In-App Browser and Image Generation
The in-app browser gives Codex access to live web content without requiring the user to copy and paste information. Codex can now research a topic, synthesize what it finds, and incorporate that into a workflow without leaving the application. Image generation — powered by DALL-E 4 or its successor — means Codex can produce charts, diagrams, mockups, or visual content as part of a workflow output. For knowledge workers producing presentations, reports, or marketing materials, this eliminates a context switch to a separate image tool.
Automation Memory
Automation memory allows Codex to remember the steps of workflows it has executed and replay or modify them in future sessions. This is a meaningful usability improvement for repetitive professional tasks — daily report generation, weekly data pulls, recurring email workflows — that would otherwise need to be re-specified each session. Memory also allows Codex to learn a user's preferences and apply them automatically, reducing the amount of instruction required over time.
Rate Limit Doubling and Pricing Signal
Doubling rate limits across paid plans is both a capacity signal and a competitive move. OpenAI is signaling that it has the infrastructure to support heavier Codex usage, and it removes one of the practical barriers to making Codex a primary workflow tool rather than an occasional supplement. The rate limit increase also responds directly to complaints from power users who had been hitting caps on intensive agentic workflows.
The Bottom Line
The Codex desktop update is OpenAI's most direct move yet into the agentic productivity software market. By adding computer control, browsing, image generation, and memory, OpenAI is building toward a product where users describe what they need and Codex handles the full execution — not just the code. The competition with Anthropic's Claude, Microsoft Copilot, and emerging agentic startups is now squarely in this territory.
Related Articles
- Anthropic Redesigns Claude Code With New Desktop App and Auto-Compact
- OpenAI Updates Agents SDK With Native Sandboxing
- Adobe Unveils Firefly AI Assistant for Multi-Step Creative Cloud Tasks