OpenAI launches GPT-5.6 Sol, Terra, Luna with White House approval

OpenAI has officially announced the release of GPT-5.6, its next-generation flagship model, alongside two lighter variants: Terra (for everyday tasks) and Luna (lower-cost). The flagship model, GPT-5.6 Sol, will roll out gradually starting July 9, following delays imposed by the White House. In a blog post, OpenAI stated it spent weeks stress-testing Sol for weaknesses, strengthening protections for high-risk activities, sensitive cyber requests, and repeated misuse. The company claims Sol is its most capable model for cybersecurity and outperforms competitor Anthropic's Claude Mythos 5 while using just one-third of the output tokens.

The launch comes amid heightened government scrutiny of frontier AI models. Reports indicate the White House initially limited GPT-5.6's release to a small group of trusted partners whose participation was shared with the government. OpenAI has now received clearance for public release. The announcement follows news that OpenAI reportedly proposed giving the U.S. government a 5% equity stake, with CEO Sam Altman in early talks with President Donald Trump. Meanwhile, competitor Anthropic faced similar restrictions, with the White House ordering it to suspend access to its Fable 5 and Mythos 5 models by foreign nationals; Mythos 5 is now allowed for U.S. critical infrastructure organizations.