OpenAI confirmed on Friday, June 27, 2026, that it is delaying the public release of its next generation of AI models, GPT-5.6, at the request of the Trump administration. The company will first share the models with a small set of government-preapproved customers before working with the administration to gradually expand access.
GPT-5.6 will be offered in three versions: Sol, the most capable tier; Terra, a mid-tier option; and Luna, a faster and more affordable version. OpenAI describes Sol as its most capable model to date on benchmarks covering cybersecurity, biology, and agentic abilities. The company says it has also developed a “layered safeguard stack” designed to prevent misuse for cyberattacks and other malicious purposes.
OpenAI is not pleased with the arrangement, according to a person familiar with the company’s thinking, but views the delay as temporary. In a blog post, the company said it hopes to make GPT-5.6 broadly available within weeks. “We don’t believe this kind of government access process should become the long-term default,” OpenAI wrote. “It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them.”
The delay follows an executive order signed by President Trump earlier this month, which called for a voluntary process allowing the government to review AI models 30 days before a wider release. However, OpenAI executives said no such voluntary framework has actually been established yet, leaving AI labs in an uncertain interim period. The company said it sends the government a customer list and receives feedback, but cannot disclose further details about how approvals are made.
The situation mirrors a recent clash involving Anthropic, which was directed to take its most advanced AI models offline following an export control directive from the White House. That dispute remains unresolved, with some Anthropic employees still barred from accessing the company’s most advanced models.
The administration’s interventions with both OpenAI and Anthropic may create uncertainty for other US AI labs, coming at a time when the White House has expressed growing concern about the cybersecurity implications of advanced AI models. OpenAI plans to expand its approved customer list next week to include some international partners.
Source: WIRED