OpenAI will begin making the GPT-5.6 model publicly available on Thursday, following a delay of several weeks due to additional security tests required by the US administration. The launch itself demonstrates that the development of the most advanced AI models is increasingly becoming a key element of national security policy.
The new family of models includes the flagship GPT-5.6 Sol, as well as the more affordable Terra and Luna variants. OpenAI states that Sol is the company’s most advanced model, whilst Terra is intended to offer a balance between performance and cost, and Luna has been designed for tasks requiring large scale and low operating costs.
Initially, access to GPT-5.6 was granted only to a small group of verified partners, whose involvement had been reported to the US authorities. According to Axios, the Donald Trump administration only agreed to a wider roll-out following additional testing and a series of meetings with OpenAI representatives.
The decision is part of a wider trend. Washington is scrutinising so-called large language models ever more closely, fearing their use in sophisticated cyberattacks or for military applications by countries such as China or Russia. At the same time, China is exploring the possibility of restricting foreign access to its own most advanced AI models, which shows that control over artificial intelligence is becoming a factor in geopolitical rivalry.
Anthropic has recently been affected by similar restrictions. The company temporarily disabled the Mythos 5 and Fable 5 models following a decision by the US authorities regarding export controls. Some of the restrictions were later lifted after additional safeguards were put in place, though the Mythos model remains available only to selected, trusted organisations.
Meanwhile, the competition shows no sign of slowing down. Elon Musk has announced that his company is making the Grok 4.5 model publicly available, which further intensifies the rivalry between the largest providers of generative AI.
The most important implication for business, however, goes beyond the launch of GPT-5.6 itself. Increasingly advanced models will not be brought to market solely when they are technologically ready. The timing of their release will increasingly be determined by national security considerations as well, meaning that regulation and cooperation with government bodies are becoming a new element in the AI development cycle.
