On Friday, OpenAI announced the introduction of three new artificial intelligence models—GPT-5.6 Sol, Terra, and Luna—while complying with a U.S. government request to restrict initial access to a small group of trusted partners. OpenAI emphasized its commitment to broad access and plans to make these models generally available in the coming weeks.
The company previewed the models' capabilities to the government prior to the launch, expressing concerns that such government access processes could hinder the availability of advanced tools to users and developers. This announcement follows a similar situation with rival Anthropic, which had to disable access to its latest models due to compliance with export control directives.
The Trump administration's recent executive order on AI regulation has prompted developers to allow government assessments of model capabilities before full releases. OpenAI is collaborating with the administration to establish a framework for these assessments.
Among the new models, Sol is highlighted as the most capable, particularly in coding, biology, and cybersecurity, although it remains below OpenAI's critical cybersecurity risk threshold. This regulatory environment may shape the future of AI development and competition in the sector