OpenAI will roll out its GPT 5.6 model in phases at the request of the Trump administration, granting access first to a small set of U.S.-based partners vetted alongside federal officials. The move, coordinated with the Office of the National Cyber Director and the Office of Science and Technology Policy, comes as the White House pursues a model-vetting framework under a new executive order. CEO Sam Altman told employees the government would approve customers “case by case” during the preview, with a broader release possible in weeks if the process runs smoothly. The suite includes three tiers—Sol, Terra and Luna—with OpenAI asserting Sol does not cross its “cyber critical threshold” and is more effective at defense than end-to-end attacks. The strategy echoes rival Anthropic’s curtailed rollout of Mythos amid federal concerns over powerful cyber capabilities, underscoring Washington’s sharper posture on advanced AI releases.
Related article:
What did we learn from Elon Musk and Sam Altman’s courtroom drama?
