OpenAI CEO Sam Altman speaks to journalists after meeting with U.S. House Minority Leader Hakeem Jeffries at the Capitol in Washington, DC, June 3, 2026.
Brendan Smialowski AFP | Getty Images
OpenAI on Friday announced three new artificial intelligence models and said it would initially comply with a U.S. government request to limit deployment to a “small group of trusted partners.”
The company said in a blog post that it “believes in broad access” and is working to make GPT-5.6 Sol, Terra and Luna models generally available in the coming weeks. OpenAI said it previewed the model’s capabilities and shared its plans with the government ahead of Friday’s launch.
“We believe this type of government access process should not become the long-term default,” OpenAI said in a statement. “We secure the best tools from the users, developers, enterprises, cyber defenders, and global partners who need them.”
OpenAI has not disclosed the names of partners who will be able to use the new model.
The announcement comes two weeks after rival Anthropic announced it needed to revoke access to two of its newest models to comply with export control directives from the Trump administration. Anthropic is actively negotiating with officials in Washington, D.C., but has not said when it expects the model to come back online.
Since President Donald Trump signed the AI Executive Order earlier this month, the Trump administration has taken a markedly pragmatic approach to AI regulation. The order provided few specific details and asked AI developers to voluntarily allow the government to evaluate the functionality of their models before full release.
OpenAI said it is working with the Trump administration to help establish a framework for such assessments and develop a “repeatable process for future model releases.”
“We are taking this short-term step because we believe this is the strongest path to achieving broader availability in the coming weeks,” OpenAI said.
GPT-5.6 Sol, Terra, and Luna are named according to their functional layers. OpenAI said Sol is its most powerful product to date.
The model shows improvements across coding and biology, and OpenAI said it is its most capable model for cybersecurity. The company said it is better at helping users remediate vulnerabilities than conducting end-to-end attacks, and has not yet crossed OpenAI’s “severe” cybersecurity risk threshold, defined as posing “unprecedented new paths to serious harm.”
Attention: OpenAI plans to restrict deployment of GPT 5.6 after Trump administration call
