The Trump administration has approved the wide public release of OpenAI's advanced GPT-5.6 model family, a source familiar with the discussions confirmed to Axios on Tuesday. OpenAI announced late Tuesday night that its flagship model, named Sol, along with the more accessible Terra and Luna variants, will launch publicly this Thursday.Prompt via GPTcommandsThe decision marks the latest delayed rollout due to coordination between the U.S. government and leading AI companies over access to frontier systems.GPT-5.6 Sol, along with Terra and Luna, will launch publicly this Thursday.
We’re expanding preview access globally now. pic.twitter.com/Uk5HcfSc2e
— OpenAI (@OpenAI) July 8, 2026GPT-5.6 sol launches thursday!
happy building
— Sam Altman (@sama) July 8, 2026OpenAI announced the models in June - with an initial commitment to allow a select group of organizations access whose "participation has been shared with the government," according to a blog post. According to the company, "we’re introducing a new max reasoning effort to give Sol the most time to reason deeply. Additionally, we’re introducing a new ultra mode that goes beyond the capabilities of a single agent by leveraging subagents to accelerate complex work."Last month, the administration directed OpenAI to begin with a limited release of GPT-5.6, restricting early access to government-approved entities only. OpenAI had publicly stated at the time that a staggered approach was not its preferred method and that both companies and regulators were operating without finalized standards called for in President Trump's recent AI executive order.The new green light followed additional testing and meetings - with technical experts from OpenAI traveling to Washington, D.C. to answer questions during the review process. The evaluation was conducted by the Center for AI Standards and Innovation (CAISI) within the Department of Commerce - the entity responsible for assessing advanced AI systems for safety, security, and standards alignment.The New NormalPowerful AI models are no longer released solely at the discretion of their creators. The U.S. government and top AI labs are actively negotiating - model by model, in real time - who gets access and under what conditions due to concerns over national security, potential misuse, the need to maintain American leadership in AI while managing downsides. Of course, big brother is also shackling US models while cheaper, more efficient, open-weighted Chinese models are starting to dominate. That said - China is now considering restricting access to their models.we distilled 2.3M Claude Fable 5 reasoning traces into Qwen3-4B
