The GPT-5.6 received approval from the US government to be released publicly. It consists of three variants – Sun, Earth and Luna – and will be available to all users from Thursday, July 9.
© Getty Images
by
A OpenAI announced that it will officially launch the GPT-5.6, its new model of Artificial Intelligence this Thursday, July 9. This new model has a total of three variants: Sun, Earth and Luna.
Of this trio, the GPT-5.6 Sun is the most powerful, while the Earth is more modest and, despite having a performance similar to GPT-5.5, was created for everyday use. Luna, on the other hand, is pointed out as the company’s most affordable model (from the cost point of view).
It is important to remember that, before being released publicly, these three variants of GPT-5.6 had to be delivered to the US government for preliminary testing. The order signed by Trump in early June required that the AI companies give the U.S. government access to their most powerful models for a period of 30 days before they were released publicly.
At the time, OpenAI stated that “this type of government access process should not be the long-term standard”, nevertheless deciding to obey to ensure a free launch of GPT-5.6.
Axios says that after multiple tests have been conducted with the help of specialized technicians from OpenAI, the division of the Department of Commerce allocated to these evaluations will have decided to give “green light” to the public launch of the GPT-5.6 and has now arrived on July 9.
GPT-5.6 Sol, along with Terra and Luna, will launch publicly this Thursday.
We’re expanding preview access globally now. pic.twitter. with/Uk5HcfSc2e
— OpenAI (@OpenAI) July 8, 2026
USA raises restrictions on the most powerful AIs in Anthropic
Washington lifted the restrictions that had forced Anthropic to block access to the two most powerful models, announced the leading U.S. artificial intelligence (IA) company, with access to have been restored in early July.
” We received notification that the Department of Commerce lifted export controls on Claude Fable 5 and Mythos 5″, imposed on behalf of national security, declared Anthropic on the social network X on Tuesday.
” We will begin to restore access,” he added.
On June 12, the U.S. Government abruptly forced Anthropic to block access to these two cutting-edge models, considering that failures had been detected in security measures aimed at preventing misuse for cyberattack purposes.
This forced withdrawal, an unprecedented situation by a government, sparked a wave of worldwide criticism, rekindling the debates on the digital sovereignty of countries dependent on US technologies.
After fierce negotiations in Washington, about which few details were released, access to Mythos 5 was unlocked on Friday for a small group of “ciberdefensors and infrastructure operators”, but only Americans.
Foreign partners, including state cybersecurity agencies in Europe and Asia, were still without access at this stage. Anthropic, with turbulent relations with the Trump Administration, did not clarify whether the lifting of export control implied the reintegration of these non-American partners.
The decision allowed the online return of Fable 5, a version aimed at the general public of Mythos, with cybersecurity restrictions and risks of biological and chemical attacks.
OpenAI, Anthropic’s American rival, launched on Friday the company’s most powerful model to date, the GPT-5.6, initially with limited access, accepting, for the first time, that the US Government validates, customer to client, the authorized partners.
For a long time hostile to any AI regulation, accused of being an obstacle to innovation against China, the Trump administration initiated a radical twist, taking into account the powerful capabilities of cutting-edge models.
Recent decisions in this area, taken in a still unclear and controversial legal framework, mark a resumption of government control over this crucial technology.
On Tuesday, the director of the Central Intelligence Agency (CIA), John Ratcliffe, compared the capabilities of the most advanced AI models to “digital nuclear weapons”, in a rare public intervention in Washington.