OpenAI has initiated a restricted preview of its latest GPT-5.6 series, comprising the Sol, Terra, and Luna models, following a direct request from the United States government. The company disclosed that the initial rollout is confined to a small cohort of vetted partners, with participant lists shared alongside the Department of Defense. This controlled release operates under an existing agreement permitting Pentagon access to OpenAI artificial intelligence infrastructure, with a broader public launch slated for the coming weeks. The measured deployment arrives amid heightened scrutiny of advanced AI capabilities and federal oversight mechanisms. Earlier this month, competitor Anthropic voluntarily revoked public access to its Mythos and Fable models after the Trump administration implemented sweeping export controls. Anthropic April preview of the Mythos model had triggered significant alarm among investors and policymakers when it demonstrated an unprecedented ability to identify software vulnerabilities and compromise cyber infrastructure. In response to such developments, the Trump administration issued an executive order on June 2 that establishes a voluntary thirty-day review process for next-generation models, while simultaneously cautioning against heavy-handed federal regulation that could cede the global generative AI race to China. OpenAI acknowledged the necessity of the current phased approach but emphasized that government-mandated access protocols should not become an industry standard. In a public statement, the company warned that prolonged regulatory gatekeeping could restrict critical AI tools from developers, enterprises, cybersecurity professionals, and international allies. As the federal landscape continues to evolve, technology firms are balancing rapid innovation with national security directives, a dynamic that will likely dictate the rollout cadence and accessibility of subsequent frontier AI models.