Read, Watch or Listen

San Francisco-based OpenAI on Thursday released its GPT-5.6 frontier model family for general availability and introduced ChatGPT Work, an agentic AI tool designed to execute complex professional tasks across applications and files. The GPT-5.6 lineup consists of three models: Sol, the flagship model that OpenAI says delivers significant performance gains in coding, biology, and cybersecurity; Terra, aimed at day-to-day workplace activities; and Luna, the fastest and most affordable option intended for routine tasks. The company said the Sol model achieved a 54% improvement in token efficiency for agentic programming tasks and performs as well as or better than other competitive models currently on the market. The GPT-5.6 models were first announced last month but were initially restricted to a limited group of trusted partners at the request of the U.S. government, while OpenAI carried out further testing and consultations with authorities before moving to a full public release. ChatGPT Work, powered by GPT-5.6, can draw context from apps, files, and workflows to produce finished documents, spreadsheets, presentations, reports, and websites, and can operate for hours on complex projects. OpenAI, which is preparing for an initial public offering, also launched a ChatGPT desktop application and a hosted websites feature, and said ChatGPT Work is rolling out on web and mobile to Pro, Enterprise, and Edu customers, with Plus and Business users to follow in the coming days. The company stated it does not want early government-only access to become a long-term default, arguing it limits access to top-tier tools for users and developers.

Prepared by Jonathan Pierce and reviewed by editorial team.