GPT-5.6 Set for Public Release After Delays by OpenAI

This week, OpenAI is set to introduce its most advanced model to date, GPT-5.6 Sol, following delays imposed by the White House.

The announcement was made by the company on X on Wednesday, revealing that the flagship model will be gradually released starting July 9. Alongside it, two additional models will be launched: the versatile Terra and the budget-friendly Luna.

This Tweet is currently unavailable. It might be loading or has been removed.

OpenAI stated in a blog post that it has dedicated weeks to rigorously testing GPT‑5.6 Sol for potential vulnerabilities. During this process, the company enhanced safeguards against high-risk activities, sensitive cyber requests, and repeated misuse. They described it as “the most capable model yet for cybersecurity,” asserting that GPT‑5.6 Sol outperforms competitor Anthropic’s Claude Mythos 5 while utilizing only a third of the output tokens—units that AI models use to formulate responses.

The announcement follows reports indicating that the White House requested OpenAI to restrict the initial rollout of GPT-5.6 to a limited testing group. OpenAI referred to this as a “small group of trusted partners whose involvement has been communicated to the government.” With recent developments, OpenAI has now received approval for the public release of GPT-5.6. This news comes shortly after OpenAI reportedly proposed granting the U.S. government a five percent stake in the company. CEO Sam Altman has been in preliminary discussions with President Donald Trump, according to the Financial Times.

In parallel, competitor Anthropic faced similar restrictions from the U.S. government. The White House mandated that Anthropic “suspend all access to Fable 5 and Mythos 5 for any foreign nationals, both inside and outside the United States.” Currently, Anthropic is permitted to share Mythos 5 with a select group of U.S. organizations responsible for operating and safeguarding critical infrastructure, while Fable 5, a variant of Mythos, is available globally.

Disclosure: Ziff Davis, the parent company of Mashable, filed a lawsuit against OpenAI in April 2025, claiming that OpenAI infringed on Ziff Davis copyrights while training and operating its AI systems.