OpenAI's GPT-5.6 Sol, Terra, and Luna go public Thursday after US lifts curbs

OpenAI is opening its GPT-5.6 model family to the public on Thursday, July 9, ending a roughly two-week period in which access was restricted to about 20 vetted partner organizations at the U.S. government’s request. The lineup spans three tiers: Sol, the flagship aimed at the hardest problems like complex coding and security research; Terra, a mid-range model tuned for high-volume business work such as support, internal tooling, and document analysis; and Luna, a low-cost option for summarization, drafting, and routine automation. Pricing per million tokens runs $5/$30 (input/output) for Sol, $2.50/$15 for Terra, and $1/$6 for Luna.

The rollout is as much a policy story as a product one. After OpenAI first shipped GPT-5.6 on June 26 under government-imposed limits, the Department of Commerce cleared a broad release following additional testing and interagency meetings — an unusual instance of a frontier model launch being gated by federal review. OpenAI has publicly signaled that it doesn’t want such restrictions to become routine.

For a technical audience, the notable details are Sol’s pitch toward cybersecurity and biology work paired with what OpenAI calls its most robust safety stack yet, including hardened protections against sensitive cyber requests and repeated misuse. The release also introduces a new maximum reasoning-effort setting and an ‘ultra’ mode that spins up subagents to parallelize complex tasks — a nod to the dual-use tension of shipping stronger agentic and security capabilities to everyone at once.