OpenAI staggers AI model release after Trump administration request

OpenAI is staggering the release of its GPT 5.6 AI model after a request from the Trump administration, launching it initially only to a small group of U.S.-based trusted partners whose participation has been shared with the federal government. The three-model series—Sol, Terra, and Luna—will expand to foreign partners next week, with the company expressing dissatisfaction that the limited rollout delays access for users and cyber defenders globally.

The decision parallels Anthropic's experience with its Mythos model, which the U.S. government ordered restricted from foreign nationals due to its cyber-hacking capabilities. OpenAI said GPT 5.6 Sol, its strongest model yet, does not cross a "cyber critical threshold" and is better at helping find vulnerabilities than executing attacks. The company is working with the White House on a vetting framework under Trump's executive order, hoping broader availability will follow within weeks if the government approval process succeeds.