OpenAI has unveiled its latest artificial intelligence models, the GPT-5.6 family, but access is sharply limited — and for now, almost no one outside the company can actually use them.
The new lineup carries celestial names. According to The Indian Express, OpenAI named the three models after objects in the sky: Sol, the Latin word for the Sun; Terra, for Earth; and Luna, for the Moon.
What sets this release apart is not just capability but restriction. Yellow.com frames it bluntly, describing OpenAI's strongest GPT-5.6 models as arriving "behind a locked door." The Indian Express likewise notes that the models have been announced, yet the public largely cannot access them yet.
Part of the reason appears to involve government oversight. According to trendingtopics.eu, GPT-5.6 is also being restricted by the US government — a notable detail suggesting that policy and regulation, not only OpenAI's own choices, are shaping who gets to use the technology.
The models are also being positioned around security. A report shared via LinkedIn says OpenAI unveiled GPT-5.6 with advanced cybersecurity features and enhanced safety controls, pointing to an emphasis on guardrails alongside raw performance.
Taken together, the sources describe a release that is unusual in shape: a flagship AI announcement paired with deliberate limits on availability, attributed both to safety design and to US government restrictions.
Why it matters: when the most powerful version of a widely used AI system launches with locked access and government involvement, it signals that the rollout of frontier AI is increasingly governed by safety and policy decisions — not just by who builds the technology first.