TECHNOLOGY

United States limits access to OpenAI GPT-5.6 models

PUBLISHED Jul 5, 2026, 11:55 AM ET

Read, Watch or Listen

Washington, United States – The US government has restricted access to OpenAI’s newly unveiled GPT-5.6 family of models, including its most powerful system, Sol, as well as Terra, aimed at daily work, and Luna, a more affordable version. OpenAI says Sol is its most capable model to date and views it as comparable to Anthropic’s Mythos. Under the new rules, OpenAI may offer these models only to a small group of government‑approved partners, making it the second firm, after Anthropic, to face such limits at the direction of the Trump administration. The move follows earlier action that forced Anthropic to disable Mythos and a guarded version called Fable 5 for all foreign nationals, including those employed at the company. Washington, United States – The latest restrictions extend a broader pattern of US controls on cutting‑edge AI systems. In early June, the administration imposed export controls on Anthropic’s most cyber‑capable models, triggering an 18‑day worldwide shutdown for foreign users before lifting the ban on June 30 and restoring access to selected “trusted” US organisations. Officials say the measures help preserve US leadership in advanced AI, but business groups and researchers warn they could create regulatory uncertainty as rivals such as China develop frontier models like Z.ai, which some observers compare with Mythos. OpenAI criticised the US decision, arguing that the curbs keep top AI tools away from users, developers, enterprises, cyber defenders and international partners who rely on early access to drive innovation.

By Sarah Whitman | JQJO News

Timeline of Events

Early June Trump administration imposes export controls
Early June Anthropic models banned for foreign nationals
Mid June Anthropic forced worldwide model shutdown
June 30 Export controls and blackout lifted
Recently OpenAI unveils Sol, Terra, Luna
Recently US restricts access to GPT-5.6 models
Recently Trusted US organisations regain Anthropic model access
Recently Chinese Z.ai model draws capability comparisons

News Intelligence

The US limiting access to OpenAI's GPT-5.6 models could affect you. If you're a developer, researcher, or business relying on these AI tools, you may face delays or hurdles. It could also impact cyber defenders and international partners. Check if your work involves these models.