OpenAI said on Friday that it will delay a full public rollout of GPT-5.6 following a request from the U.S. government, restricting initial access to a small cohort of vetted partners. The company reported that it shared details about those partners with U.S. authorities as part of the arrangement.
OpenAI described the limited release as a temporary measure while it collaborates with Washington to develop a broader framework for future model launches. In a blog post, the company said it had briefed government officials on its plans and on the capabilities of the new models prior to the planned launch.
Government concerns and the voluntary framework
U.S. policymakers have expressed rising concern about the national security risks that could accompany increasingly capable AI systems, pressing companies to implement tighter guardrails. Part of the stated rationale for the U.S. request is to ensure officials have early visibility into potential threats - including cyberattacks and military misuse - before advanced tools are made widely available.
Earlier this month, President Donald Trump signed an executive order that established a voluntary process for AI developers. Under that guidance, developers can offer "covered frontier models" to the U.S. government for up to 30 days ahead of wider distribution to trusted partners. OpenAI said its decision to limit the initial release aligns with efforts to work with the Administration on what it called the cyber Executive Order framework and to build a repeatable process for future releases.
"We are taking this short-term step because we believe it is the strongest path to broader availability in the coming weeks, while we work with the Administration to develop the cyber Executive Order framework and a repeatable process for future model releases," OpenAI said.
OpenAI's testing, partner coordination and concerns about permanence
The company said it will continue intensive testing and close coordination with the vetted partners it has identified as it prepares for a subsequent, wider release. OpenAI also cautioned that the level of government access and oversight being used for this limited rollout should not become a permanent standard.
OpenAI did not publicly disclose the names of the partners that will receive early access. The company also warned that a process like the one being used could restrict access to advanced AI capabilities for users that include developers, businesses, cybersecurity professionals and international partners who might benefit from earlier access.
The model lineup
At the center of the new offering is GPT-5.6 Sol, which OpenAI described as its most advanced model to date. The company is also making available mid-tier Terra and a lower-cost model, Luna, as part of the new lineup.
Context within markets and trading references
The announcement included material that referenced trading interest in OAI. Separately, promotional text in the original circulation discussed chart analysis tools and trading plans linked to OAI, though OpenAI itself focused its communication on the limited model access and cooperation with government authorities.
As OpenAI works with U.S. officials on the framework, the company intends to preserve rigorous evaluation while preparing for broader availability, but it emphasized that the temporary government access arrangement should not become a default practice.