OpenAI receives US approval for an extensive rollout of GPT-5.6 following several weeks of government testing.

      OpenAI has received approval to broadly release its most advanced model, following the US government's authorization for a wider rollout of GPT-5.6, which had been delayed for weeks due to new regulatory oversight on frontier AI in Washington. Previously, the model was accessible only through a limited preview to about 20 partners, all of which had been individually sanctioned by the US government. This arrangement, which was the first of its kind for an American frontier model, has now been superseded by the broader release.

      This approval came after additional evaluations conducted by the Commerce Department's Center for AI Standards and Innovation, the entity established to assess advanced systems. According to Axios, OpenAI dispatched technical experts to Washington to address the agency's inquiries.

      GPT-5.6 is structured as a three-tier system rather than a single model, with Sol as the flagship, Terra as a more economical mid-tier option, and Luna being the fastest and least expensive of the three. OpenAI describes Sol as being particularly adept at coding, biology, and cybersecurity, and it features a “max reasoning effort” mode that allows the model additional time to tackle complex problems. These attributes, especially those relating to biology and cybersecurity, were reasons the government sought a closer examination before the broader release.

      The tiered system represents both a commercial and technical decision. Terra targets everyday enterprise needs where cost is more critical than sheer capability, while Luna is optimized for high-volume tasks requiring speed, allowing OpenAI to set varied pricing across the same model family.

      The confined preview prior to the broader rollout was particularly stringent. For weeks, GPT-5.6 was exclusively available to a select group of organizations whose identities OpenAI had disclosed to the government, marking a first for an American lab gating a frontier model behind a government-approved list.

      This review process is part of a framework established by the Trump administration on June 2, which introduced a voluntary pre-release vetting for the most capable models. The case of GPT-5.6 extended beyond voluntary review to a government-managed access list, a measure OpenAI agreed to only after being asked to postpone the launch.

      OpenAI has expressed its discomfort with establishing such a precedent. The company has stated it does not believe this kind of government access procedure should be the standard moving forward, despite agreeing to participate this time.

      This unease is understandable. A government capable of controlling a launch can just as easily halt one, as evidenced by the administration's past actions in the sector, such as directing Anthropic to discontinue two models.

      For OpenAI, the commercial implications of the delay were significant. Each week that GPT-5.6 remained within a limited partner preview was a week competitors could attract the enterprise customers OpenAI aimed to reach with the new tiers.

      The company anticipates expanding access to GPT-5.6 within days, building on the foundation laid with GPT-5.5 earlier this year. OpenAI has indicated that all three tiers will be available to the public in the upcoming weeks, although a specific date has not been announced.

      OpenAI is not the only laboratory subject to the new regulatory framework. The same oversight applies to its competitors, suggesting the manner of this rollout may influence how other US companies will introduce their next frontier models to the market.

      This episode establishes not so much a timeline for the product as a new model for release. For the first time, a leading US laboratory has rolled out a frontier model on the government's timetable instead of its own, and now both parties must determine whether this will be an isolated occurrence or a precedent for future releases.