OpenAI Launches GPT-5.6 But Limits Access Under Trump Administration Pressure

OpenAI has introduced its latest GPT-5.6 model series but immediately restricted access to a small group of trusted partners, following directives from the Trump administration. The company announced on June 28, 2026, that early access is granted only to partners approved by the U.S. government. OpenAI emphasized that such government-controlled access should not become the standard, as it excludes many users, developers, cybersecurity defenders, and global partners who need these advanced tools.

The GPT-5.6 series includes three models: Sol, the flagship and most powerful model; Terra, a more balanced model for everyday use; and Luna, a faster and cheaper option. Despite Sol’s superior capabilities, the Trump administration has limited the launch of all three models. The names reflect their scale, meaning sun, earth, and moon respectively. This move comes amid increased U.S. government pressure on AI companies to restrict access to their most advanced systems. For example, after Anthropic released its Fable 5 model, the government ordered the company to block access for foreign users, leading to the model’s complete withdrawal.

Former White House AI advisor Dean Ball, who is expected to join OpenAI, criticized the administration’s recent executive order requiring voluntary pre-launch government review of advanced AI models within 30 days. He described it as an involuntary licensing regime causing heavy restrictions and potential indefinite delays, which could benefit China in the AI race and risk billions invested in AI infrastructure. OpenAI said it is working with the government on a new cybersecurity executive order framework and a recurring review process for future model launches.

OpenAI highlighted that GPT-5.6 Sol offers enhanced agentic capabilities in coding, biology, and cybersecurity, with advanced "max" and "ultra" thinking modes using coordinated sub-agents for complex tasks. It reportedly outperforms Anthropic’s Mythos 5 in coding workflows while using only a third of the tokens. Sol also features OpenAI’s strongest security measures yet, designed to resist adversarial attacks and prioritize defensive cybersecurity over exploiting vulnerabilities. Safety mechanisms are integrated into the core model behavior rather than relying on separate filters, addressing issues Anthropic faced with Fable 5.

Initially, GPT-5.6 models are available only to a select group, but OpenAI plans to expand access soon to ChatGPT, Codex, and API users. Pricing varies by model size: Sol costs $5 per million input tokens and $30 per million output tokens; Terra is half that; Luna costs $1 and $6 respectively. Input tokens are user prompts, while output tokens are the model’s generated responses. OpenAI also improved prompt caching to reduce costs and improve predictability for repeated use.