OpenAI launches GPT 5.6, after which it will be restricted to U.S. partners

• OpenAI is launching GPT 5.6 in stages for an existing group of U.S. partners.
• The company plans to gradually comply with the content requirements, and the strictest limitations will apply to the first stage.
• The context suggests that the restrictions on the rollout to the U.S. are related to the Mythos and Fable discussion, which involved export controls.

For more details, see The Verge and The Verge.

OpenAI says it is launching its GPT 5.6 "ineligible for the group" model for U.S. partners. The company, however, notes that it is not sharing the full details of the plan, mentioning only that it will be implemented gradually.

In other words, OpenAI says that its model will start rolling out to U.S. partners, but only for those partners that comply with the content requirements. In addition, OpenAI says that it will not be available in the U.S. for all partners, but only for those that meet the requirements.

Meanwhile, the company is also facing restrictions on the export of models. Specifically, it is required to comply with the "advanced computing" restrictions, which limit the use of models. According to OpenAI, the company is complying with these requirements and will limit the use of its models to the U.S.

GPT 5.6 is scheduled to be released in three versions: Sol, Terra, and Luna. OpenAI has described GPT 5.6 Sol as the "most powerful" model on the user side, and also noted that it will not be possible to use the "internally developed" model in its system without restrictions. The company says Sol is designed to help with data processing, but not to bypass the restrictions.

As for the material, the general director of OpenAI, Sam Altman, said that the federal government will be able to process the model. In turn, he said that the company will be able to use the "content" model, and the strictest release will be "as soon as possible", which is already in progress.

Meanwhile, the situation with Anthropic is also expected to be similar to the Mythos and Fable models. The material also says that Anthropic has suspended or delayed the rollout of its models to the U.S. due to the restrictions that would be imposed on the company’s partners. However, it is noted that the company is taking steps to ensure that the restrictions do not affect the release of the model. It is also noted that the exact details of the restrictions have not been disclosed.

The text also states that the U.S. is working on a new framework for export controls, which would allow the federal government to regulate the use of models. The material adds that the new framework is expected to be implemented soon, and that the U.S. may impose additional restrictions on the release of models.

Tags: USA/Politics/Technology/AI