OpenAI's ChatGPT Work: Autonomous Agent Builds Apps

OpenAI has launched ChatGPT Work, an autonomous agent that can perform tasks like research, document creation, and web application building for hours without constant human oversight. This new agent runs on GPT-5.6, the company's latest model family, also released today. ChatGPT Work handles multi-step tasks, such as monitoring apps, running scheduled jobs, and producing deliverables. Users can track progress, adjust directions, or approve actions as it works. It functions across web, mobile, and desktop, accessing local files, a built-in browser, and third-party plugins. OpenAI is also merging its Codex developer tool directly into the ChatGPT desktop app for macOS and Windows, creating a unified app that combines Chat, Work, and Codex. A new Sites beta feature lets users turn work into interactive websites or web apps within ChatGPT. ChatGPT Work is available now for Pro, Enterprise, and Edu users, with Plus and Business subscribers gaining access soon. Free and Go users are excluded. The GPT-5.6 lineup includes three tiers: Sol, Terra, and Luna. Sol is the flagship model, with Terra and Luna as more budget-friendly options. OpenAI claims Sol delivers state-of-the-art results in areas like coding and cybersecurity. An ultra acceleration mode is available for Pro and Enterprise users in ChatGPT Work, allowing multiple agents to coordinate on parallel tasks. OpenAI will retire GPT-5.4 on July 23rd, while GPT-5.5 models will remain available. This development could significantly change how many businesses manage complex digital tasks.