OpenAI postpones full launch of its new models in response to a request from the White House.

OpenAI plans a phased rollout of its next-generation artificial intelligence (AI) models, following a request from the US administration to initially make the new models available to a limited number of users. This move reflects growing government concern about the potential risks of advanced AI technologies. According to the company, the new GPT-5.6 family of models, which includes Sol , Terra , and Luna , will first be made available to a select group of users before a wider release in the coming weeks.

OpenAI stressed in an official statement that government review should not become a permanent policy, noting that delaying the availability of models affects developers, companies, researchers, and cybersecurity teams that rely on the latest artificial intelligence technologies in their work.

This trend comes at a time when AI companies are becoming more cautious about releasing their most advanced models, amid growing concerns about cybersecurity and the potential for misuse of these technologies. OpenAI is not alone in adopting this approach; Anthropic previously implemented a similar policy after launching its advanced model, Cloud Fable 5 , before later suspending the availability of both Fable 5 and Mythos 5 due to US restrictions related to national security and export controls on AI technologies.

This shift indicates a clear change in the mechanism for deploying artificial intelligence models, as the latest innovations are no longer expected to move directly from development labs to millions of users, but will be launched in stages to allow sufficient time for testing and assessment of their potential risks by developers, regulators and security experts.

OpenAI’s caution isn’t limited to launching new products; it extends to the company’s plans for a stock market debut. According to a report published by The New York Times last week, the company is considering postponing its initial public offering (IPO) until 2027, following the significant volatility experienced by SpaceX after its initial public offering.

OpenAI’s market capitalization is currently estimated at over $850 billion , making it one of the largest IPOs in US history. Anthropic is approaching these levels, with a valuation of approximately $965 billion .

Observers believe that postponing the IPO may give OpenAI a better chance of listing in a more stable market environment, especially with the increasing scrutiny that giant companies with huge valuations are facing from investors and regulators.