OpenAI said it will initially roll out three new AI models—GPT-5.6 Sol, Terra and Luna—only to a small group of “trusted partners” after a request from the U.S. government. The company framed the restriction as a short-term step while it works with the Trump a
By Friday, OpenAI was ready to put three new models into the world. But the launch came with a gate—one the company said was built at the government’s request.
OpenAI announced that its next wave of artificial intelligence. including GPT-5.6 Sol. Terra and Luna. will be limited at first to a “small group of trusted partners.” The company said it is complying with the U.S. government’s request as it previews the models’ capabilities before pushing for broader availability in the coming weeks.
In a blog post. OpenAI said it “believes in broad access” and that it shared its plans and model previews with the government ahead of Friday’s launch. The company argued against turning access controls into a long-term default. saying the process would keep “the best tools” from the users. developers. enterprises. cyber defenders. and global partners that. in its view. need them.
OpenAI did not disclose which partners are included, nor did it provide names or numbers for the group.
The rollout limit lands at a tense moment for AI companies navigating Washington’s shifting approach. It comes two weeks after rival Anthropic said it had to disable access to two of its latest models to comply with an export control directive from the Trump administration. Anthropic said it was in active negotiations with officials in Washington. D.C. but it had not said when it expects the models to come back online.
The Trump administration has taken a more hands-on stance toward AI regulation since President Donald Trump signed an AI executive order earlier this month. The order, OpenAI said, was thin on specific details. It asked AI developers to voluntarily allow the government to assess model capabilities ahead of a full release.
OpenAI said it is working with the Trump administration to help establish a framework for those assessments and to develop a “repeatable process for future model releases.” The company said it is taking what it called a short-term step because it believes it is the strongest path to broader availability in the coming weeks.
The company also described the capabilities of the three models. GPT-5.6 Sol, Terra and Luna are named according to capability tiers, and OpenAI said Sol is its strongest offering yet. It said the models show improvements across coding and biology. and that GPT-5.6 Sol is its most capable model for cybersecurity.
OpenAI said the model is better at helping users fix vulnerabilities than carrying out end-to-end attacks. The company also said it still does not cross into OpenAI’s “critical” cybersecurity risk threshold, which it defined as bringing “unprecedented new pathways to severe harm.”
All three points—who gets early access, how the models are being evaluated by the government, and what risk boundaries OpenAI says it has not crossed—sit side by side in the company’s explanation for why Friday’s launch doesn’t look like a full release.
For now, the public will have to wait while a limited group of “trusted partners” gets the models first. OpenAI says the broader rollout is coming “in the coming weeks,” but the company’s message makes one thing clear: the path to availability is being shaped by Washington’s assessment process.
OpenAI GPT-5.6 Sol Terra Luna AI regulation Trump administration AI executive order trusted partners cybersecurity models export control Anthropic