OpenAI, the developer of ChatGPT, announced that it will delay the full public launch of its latest series of artificial intelligence models, GPT-5.6, in response to a request from the U.S. government. Access to these models will be limited to a small group of trusted partners during the initial phase.

In a blog post published alongside the announcement, the company stated that it had begun a "limited preview" of the GPT-5.6 series, which includes three models named Sol, Terra, and Luna.

OpenAI said it reviewed the launch plans and model capabilities with the U.S. government before making them public. The company also began, at the request of the government, providing access to these models to a small group of trusted partners who were informed by the government about their participation, before wider release.

OpenAI confirmed that it intends to make the models available to all users in the coming weeks. It noted that the current restrictions are temporary measures being implemented alongside work with U.S. authorities on establishing a permanent mechanism for security reviews of advanced AI models.

The company emphasized that it does not see giving government access to the models as something that should become a permanent practice, considering such an approach could deprive users, developers, and cybersecurity experts worldwide from benefiting from the latest advancements in artificial intelligence technology.