OpenAI said on Friday it is restricting the release of Gpt-5.6 Sol to a small group of trusted partners approved by the Trump administration, turning a new AI launch into a managed test run instead of a broad rollout.
The company said the model will stay in that limited phase for now, even as it sees the move as a temporary step on the path to broader availability in the coming weeks. For readers searching now, that means the model is not being opened to the public at launch, and the timing of wider access depends on how the review proceeds.
The restriction matters because President Donald Trump signed an executive order earlier in June that created a federal framework for vetting the national security risks of the most advanced AI systems for up to 30 days before public release. Participation was described as voluntary, but OpenAI is now acting within that process, showing how a government review can shape when a model reaches users.
OpenAI says the model is better at helping people find and fix vulnerabilities than it is at carrying out cyberattacks, and it also says GPT-5.6 Sol does not cross the company’s own risk threshold. Even so, the company acknowledged that unforeseen risks could emerge if the model is paired with other tools, and that uncertainty is why it is coupling stronger safeguards with a phased release.
That leaves the key question on the table: the model is available only to a small group of approved partners, but OpenAI has not named them. The company said it does not want this kind of government access process to become the long-term default, yet for now the review is the gate that decides who gets in and when the rest of the market can follow.
The move comes after another AI maker took offline two new AI models just days after releasing them, underscoring how quickly advanced systems can run into policy limits once they are in public view. For now, GPT-5.6 Sol is less a launch than a controlled opening, and the next stage will depend on whether the temporary testing period stays temporary.
