OpenAI releases GPT-5.6 model suite and ChatGPT Work agent

OpenAI publicly launched its most advanced artificial intelligence model family, GPT-5.6, and a new enterprise productivity agent called ChatGPT Work on Thursday, July 9, 2026. The rollout follows a period of limited availability under U.S. government oversight and marks a significant step in the company's push to embed autonomous AI directly into professional workflows.

The GPT-5.6 family consists of three models. GPT-5.6 Sol is the flagship model, designed for complex reasoning, coding, and scientific research. It features a new "Ultra" mode that can coordinate up to four AI agents working in parallel to solve complicated tasks. The other variants are GPT-5.6 Terra, positioned as a balanced, lower-cost option, and GPT-5.6 Luna, designed for fast, affordable performance. The models are becoming available across ChatGPT, the Codex coding assistant, and the OpenAI API, with global deployment expected within 24 hours.

Powering the new capabilities is ChatGPT Work, an AI agent that can orchestrate multi-step projects across a user's applications and files for hours at a time. The tool can connect via plugins to software including Slack, Microsoft Teams, Google Drive, Salesforce,...