OpenAI will make its most advanced language model, GPT-5.6, publicly available on Thursday after postponing its launch last month at the request of the U.S. government. The delay had been prompted by national security concerns: authorities worried that high-capability AI could be misused for military or intelligence purposes by China, Russia or other states deemed risky by Washington.
The announcement follows a recent U.S. decision to lift restrictions on Anthropic’s newest models, Fable and Mythos, which had been subject to an access suspension three weeks earlier for similar security reasons.
According to Axios, the U.S. Department of Commerce approved broader deployment of GPT-5.6 after conducting additional government tests under a new oversight framework for advanced models. Prior to this wider rollout, OpenAI limited access to GPT-5.6 to a small, controlled set of partners and shared testing data with authorities.
In a Tuesday statement, OpenAI said it will launch three versions of the model: GPT-5.6 Sol, Terra and Luna. Sol is described as OpenAI’s most capable system to date; Terra is positioned as a mid-tier, more affordable option; and Luna is intended as the most cost‑efficient offering.
The stricter review process is tied in part to an executive order from President Donald Trump that established a voluntary framework under which AI developers may hand over their most advanced models to the U.S. government for testing for up to 30 days before making them available to trusted partners.
Anthropic has warned that it is ‘‘likely impossible’’ to make any AI model completely resistant to attempts to circumvent restrictions. The current approvals and oversight illustrate how U.S. agencies are using new regulatory tools to identify and mitigate potential risks posed by advanced generative AI.
Why it matters
Public release of GPT-5.6 is significant both technologically and politically: it represents OpenAI’s current top technical capability while serving as a test case for how regulators and companies handle deployment of powerful models. The three-tier rollout—Sol, Terra and Luna—aims to meet different user and commercial needs, from high-performance applications to cost-sensitive, wide-scale use.
