OpenAI is limiting the release of its next-generation GPT-5.6 model lineup to a “small group of trusted partners” following a request from the U.S. government. The company made the announcement Friday, and it’s not happy about it.
The GPT-5.6 family includes three models: Sol, the flagship; Terra, a balanced model for everyday use; and Luna, a faster, lower-cost option. All three are caught up in the restrictions. OpenAI said the preview is limited to partners “whose participation has been shared with the government.”
The move follows a pattern. The Trump administration has been putting pressure on AI companies to gate their most advanced systems. After Anthropic released its Fable 5 model, the White House ordered the company to cut off access for foreign nationals, forcing Anthropic to pull the model entirely.
Dean Ball, a former White House AI adviser who’s about to join OpenAI, has argued the administration’s executive order on AI oversight amounts to a “de facto involuntary licensing regime” for frontier models. He warns it could hand China a competitive edge and put billions of dollars in AI infrastructure spending at risk.
OpenAI’s own response was measured but pointed. A blog post stated: “We don’t believe this kind of government access process should become the long-term default. It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them.” The company called it a “short-term step” and said broader availability would follow in coming weeks.
On the technical side, OpenAI says GPT-5.6 Sol outperforms Anthropic’s Claude Mythos 5 on coding benchmarks while using fewer tokens. The model also introduces an “ultra” mode with coordinated subagents for complex tasks — the kind of feature that can send usage costs soaring.
OpenAI clearly wants to position Sol as both powerful and responsible. The company says security guardrails are baked into the model itself rather than bolted on as a separate filter, a direct lesson from the Fable 5 episode.
The bigger question this raises: who controls access to the frontier of AI capability — the companies building it, or the governments regulating it?
