When presenting "GPT-5.6" the US government imposed restrictions
Leading artificial intelligence company OpenAI has unexpectedly encountered a roadblock in releasing its newest and most powerful model generation, the GPT-5.6 family, to the public, Zamin.uz reported.
Following a directive from the U.S. government, access to this technology has been temporarily restricted to a limited number of trusted partners, escalating the debate over AI safety and state oversight to a new level.
This was reported by Techcrunch.com. The new GPT-5.6 lineup consists of three core models: Sol, the company’s most advanced flagship; Terra, balanced for everyday tasks; and Luna, a fast and cost-efficient option.
Although the Sol model achieves the highest performance metrics in OpenAI’s history, the Trump administration has temporarily blocked the broad public release of all three models. According to Ixbt.com, these restrictions are part of a broader policy to strengthen government oversight of AI systems.
When discussing the capabilities and competition of GPT-5.6, OpenAI representatives emphasized that the Sol model demonstrates unmatched results in programming, biology, and cybersecurity, featuring an ultra mode that coordinates multiple internal assistants to solve complex tasks.
The model also reportedly outperforms Anthropic’s Claude Mythos 5 in coding benchmarks while being significantly more resource-efficient. The government’s intervention, however, extends beyond OpenAI.
Shortly after Anthropic unveiled its powerful Fable 5 model, the administration ordered a restriction on foreign nationals’ access to it, ultimately forcing Anthropic to withdraw the model entirely.
Such developments are creating confusion between tech companies and the White House. As former OpenAI advisor Dean Ball noted, the government’s requirement to submit new models for review 30 days before release is effectively establishing a mandatory licensing system amid growing concerns about safety and the future.
He warned that without clear safety standards, delays in the approval process could cause the U.S. to fall behind China in the AI race and put billions of dollars in sector investments at risk.
Although OpenAI announced on Friday that it is cooperating with the government, the company did not hide its dissatisfaction with the current situation.
Company officials stressed that such restrictions must not become permanent, arguing that keeping the best tools away from users, developers, and cybersecurity defenders hinders technological progress.
Currently, OpenAI is working with the government on new cybersecurity protocols and plans to strengthen collaboration with state agencies to develop safety requirements for the GPT-5.6 model family.
Both the company and its partners agree that this process must be open and transparent. Looking ahead, it is expected that safety standards will be established to allow broad public use of these models, and solutions will be found that satisfy all parties involved.
