OpenAI Broadly Releases GPT-5.6 and ChatGPT Work Following Government Stagger Delay
On July 9, 2026, OpenAI broadly released its newest flagship model family, GPT-5.6, alongside a new autonomous productivity tool called ChatGPT Work. The release comes after a unique regulatory friction point: the Trump administration requested that OpenAI temporarily delay and stagger its rollout, prompting a "collaborative back and forth" and subsequent changes to the deployment.
GPT-5.6 is offered in three distinct model flavors optimized for different tasks:
- Sol: The most powerful model, featuring a specialized "ultra" mode that allows the system to run deeper reasoning tasks and autonomously delegate work to submodels. CEO Sam Altman stated that Sol is 54% more token-efficient on agentic coding tasks than direct rivals.
- Luna: Optimized for low latency and high-speed generation.
- Terra: Designed as a balanced model for everyday productivity and general knowledge work.
Alongside the models, OpenAI launched ChatGPT Work, an agentic workspace tool powered by GPT-5.6. It can gather context across connected applications and local files to autonomously compile documents, spreadsheets, and presentations, rolling out first to Mac and Windows desktop apps.
Public and developer reception has highlighted a strong contrast between OpenAI and Anthropic. While some testers note that Anthropic's Fable 5 possesses superior raw intelligence, GPT-5.6 is being hailed as the more reliable, practical "city car" for structured daily workflows and computer-use agent tasks.