OpenAI will publicly release its most powerful model, GPT-5.6, on Thursday after
a US request delayed last month’s planned launch on national security grounds;
the model had been available only to vetted partners and its details shared with
US authorities. Washington has tightened scrutiny of advanced model releases
amid a global race to build frontier AI and warnings these systems could
accelerate cyberattacks on complex legacy networks. ANTHROPIC disabled MYTHOS 5
and Fable 5 for all users on June 12 under a US export-control order; those
restrictions were lifted last week after ANTHROPIC implemented safeguards.
SpaceXAI said it will release Grok 4.5 the same day.