OpenAI is set to publicly launch GPT-5.6 on Thursday after a delayed release stemming from US government national security concerns. The launch includes three models: Sol (most advanced), Terra (mid-tier), and Luna (most cost-efficient). The US Department of Commerce approved the broad launch following additional government testing under a new oversight framework for frontier AI. This framework, established by a Trump executive order, requires AI developers to offer frontier models to the government for up to 30 days before wider release. The news follows the US government lifting similar curbs on Anthropic's Fable and Mythos models. Anthropic has also warned that making AI models fully robust against jailbreaks is likely impossible.
Table of contents
Read: Washington backs down on Anthropic AI export curbs56 Impressions