The Trump administration has cleared OpenAI to widely release its most advanced model, GPT 5.6, ending a weekslong preview that had confined the system to a small group of government-vetted partners and marking the second time in about a week that the Commerce Department has loosened restrictions on a frontier AI product. OpenAI is expected to open access within days.

The clearance follows additional testing by the Commerce Department's Center for AI Standards and Innovation and a series of meetings between OpenAI and federal officials, according to Axios, which first reported the decision Tuesday, citing a person familiar with the matter.

OpenAI kept a team of technical staff stationed in Washington to field questions from regulators during the review. Reuters, in its own write-up, said it could not independently verify the report.

GPT 5.6 landed under an oversight regime built on President Donald Trump's June 2 executive order, "Promoting Advanced Artificial Intelligence Innovation and Security," which created a voluntary process allowing federal reviewers up to 30 days of prerelease access to models designated as "covered frontier" systems.

The order stops short of a mandatory licensing regime but directs Treasury, Homeland Security, and the Commerce Department to establish benchmarks that determine which models trigger review.

OpenAI unveiled the GPT 5.6 family in June, with a flagship called Sol and two lower-cost siblings, Terra and Luna, but agreed at the government's request to hold back broad access.

The company said at the time that a staggered rollout was not its preferred approach and that AI companies and the federal government were operating "before more concrete standards for releasing such models" had been finalized.

Officials had flagged the model's biology, coding and cybersecurity capabilities as reasons to look closer, echoing concerns that pulled rival Anthropic's Mythos and Fable models into a similar hold last month.

The Commerce Department has now lifted restrictions on both companies within about a week, suggesting the current review process is settling into a case-by-case approach rather than a fixed set of rules.

For OpenAI, the practical effect is immediate: A model that had been available only to a short list of federally vetted customers can now be released through ChatGPT, Codex, and the API.

For Washington, GPT 5.6 becomes the second live test of a framework whose formal standards are not due until Aug. 1.