OpenAI limits – OpenAI says it is limiting preview access to its GPT-5.6 models—Sol, Terra, and Luna—at the request of the US government, starting with a small group of trusted partners before broader availability in the coming weeks. The company links the step to its earlier
When OpenAI planned to launch access to its latest GPT-5.6 models on Friday, the rollout came with a new condition: the preview is being narrowed at the request of the US government.
OpenAI said it has previewed the capabilities of its GPT-5.6 model series—GPT-5.6 Sol. GPT-5.6 Terra. and GPT-5.6 Luna—with the US government ahead of Friday’s launch. It is now moving forward with a limited preview for a small group of trusted partners. with the names of those partners shared with the government.
“At their request, we are starting with a limited preview for a small group of trusted partners whose participation has been shared with the government, before releasing more broadly,” the company said.
The models being previewed are designed for different kinds of work. OpenAI described GPT-5.6 Sol as “our next generation frontier model.” It characterized GPT-5.6 Terra as a “balanced model for efficient. everyday work. ” and GPT-5.6 Luna as “a fast and affordable model for high-volume work.” OpenAI also said the new models will become available to the wider public in the coming weeks.
OpenAI framed the limitation as a short-term compliance step tied to an earlier framework with the Defense Department. The company said it is complying as part of an agreement that allows the Pentagon to use its AI models.
But the company was careful to say it should not turn into a permanent operating model. OpenAI said such a process “shouldn’t become the long-term default.”
“It keeps the best tools from users. developers. enterprises. cyber defenders. and global partners who need them. ” OpenAI said. adding that it is taking this “short-term step” because it believes it is the strongest path to broader availability in the coming weeks. The company said it is also working with the Administration to develop the cyber Executive Order framework and a repeatable process for future model releases.
The move lands in a wider moment of heightened government scrutiny around AI access. OpenAI’s statement comes after Anthropic removed all access to its latest models. Mythos and Fable. after the government raised security concerns earlier this month and ordered the company to limit access to foreign users.
For now, OpenAI says the limited preview is temporary and tied to the timing of a broader release. The company’s message is direct: access is being narrowed for the moment. but the goal is expansion—once the government’s framework work is in place—before this kind of restriction hardens into something more enduring.
OpenAI GPT-5.6 Sol Terra Luna US government request Defense Department agreement cyber Executive Order framework AI model access Anthropic Mythos Fable security concerns foreign users
So the government basically gets to control who can use GPT now? Cool cool.
I don’t get it, they’re saying it’s temporary but also “trusted partners” which sounds like insiders only. My guess is they just don’t want hackers using it and they’re calling it compliance.
Wait so Terra is the “efficient everyday” one and they’re restricting that? That’s literally the one I would use for work emails. Also why is the Pentagon asking for this stuff like it’s a normal software license?
This sounds like the same thing as when they “rolled out” Friday and then it didn’t work for regular people. And now it’s named Sol Terra Luna like some game, but it’s still controlled by the government. Like first they remove it for foreign users, then they say it’s for cyber defenders… and then it’s just another paywall later. I’m not saying it’s bad, I’m just saying it’s never as temporary as they claim.