OpenAI Staggers GPT-5.6 Release at Trump Administration Request
6 verified5 unconfirmed
OpenAI has delayed the public release of its new GPT-5.6 model suite at the request of the Trump administration, opting for a limited preview with government-approved customers. The company unveiled three model variants—Sol, Terra, and Luna—on Friday but said broader availability would come only after a review process. The move follows a similar pattern for rival Anthropic, whose Mythos model was pulled offline after the U.S. government ordered restrictions. President Trump signed an executive order earlier this month creating a voluntary framework for vetting powerful AI models, though OpenAI executives said no such framework exists yet. The company stated it does not believe government access should become the long-term default, calling the arrangement a short-term step. The development reflects growing White House concern over cybersecurity risks from advanced AI models, a shift from the administration's previous deregulatory stance.
What’s verified
OpenAI is releasing GPT-5.6 in a limited preview with customers preapproved by the U.S. government, following a request from the Trump administration.
The model suite includes three versions: Sol (flagship), Terra (mid-tier), and Luna (fast and affordable).
Anthropic’s Mythos model was previously pulled or restricted after the U.S. government ordered the company to limit access by foreign nationals.
President Trump signed an executive order earlier in June 2026 to create a voluntary framework for government review of powerful AI models.
OpenAI said it hopes to make GPT-5.6 generally available in the coming weeks and does not view the government approval process as a permanent solution.
The White House has grown more concerned about cybersecurity capabilities of new AI models, prompting the regulatory shift.
Not yet confirmed
Whether the U.S. government’s approval process is currently voluntary or effectively mandatory, as OpenAI executives said a voluntary framework does not yet exist.
The exact role of Commerce Secretary Howard Lutnick, who according to a single source called OpenAI CEO Sam Altman to demand additional agency approvals.
Specific pricing details for the model variants, including per-token costs reported by one source.
The status of Anthropic’s internal employee access to its most advanced models, mentioned by one source.
Whether the government’s review process will be completed within the “couple of weeks” timeframe cited by Altman.
Misconceptions
Some have speculated that the government process is intended as a de facto licensing regime for AI models. The executive order includes a carve-out that says it should not become such a regime, though critics and industry observers note that the current process functions as one in practice. Sources also address the notion that regulatory scrutiny is purely a competitive tactic by one AI lab against another, with commentators arguing the issue is broader than rivalry between OpenAI and Anthropic.
Key figures
Sam Altman – CEO of OpenAI
Howard Lutnick – U.S. Commerce Secretary (mentioned in one source)
Donald Trump – U.S. President
JD Vance – U.S. Vice-President
Dean Ball – GMU fellow and incoming OpenAI employee (mentioned by one source)
Peter Steinberger – creator of OpenClaw (mentioned by one source)
Sources: The Guardian, TechCrunch, Wired, The Verge
