OpenAI limits GPT-5.6 rollout after US government request

OpenAI announced the launch of GPT-5.6, offering three model variants—Sol (high‑performance), Terra (cost‑balanced) and Luna (fast, affordable). The company said the new models would be available only through the OpenAI API and Codex to a small group of trusted partners, with no public waitlist and no access via ChatGPT during the preview period.

At the request of the United States government, OpenAI delayed broader distribution. The move follows a June 2, 2026 executive order requiring AI developers to share frontier models with federal agencies for up to 30 days before wider release. This marks the first instance of the US pre‑emptively asking an American AI firm to restrict a model launch. OpenAI cited high capability in cybersecurity and bio‑chemical risk for GPT-5.6 and said additional safeguards and monitoring have been added, though the models do not reach the highest “Critical” risk level.

Developers in the community have reported detecting GPT-5.6 in Codex sessions using a “Juice” parameter test, suggesting that some users may already be receiving the new model as part of an A/B deployment, even without official invitations.