GPT-5.6 Goes Public: Sol, Terra, Luna and the Prices

The preview ran 13 days, starting 26 June with roughly 20 vetted partner orgs under the Trump administration's June AI executive order, which asks for voluntary pre-release government review of frontier models. On 9 July OpenAI opened GPT-5.6 to everyone, staging the rollout over about 24 to 48 hours. Free-tier users keep GPT-5.5 as their default.

The company did not go quietly. OpenAI publicly objected to the review process becoming "the long-term default," arguing it keeps the best tools from users, developers, enterprises and cyber defenders. That makes the launch both a product story and the first real test of the White House's voluntary framework.

Three tiers. Sol is the frontier model, with a heavier Sol Ultra mode that works harder and delegates to sub-models. Terra hits ~5.5-level intelligence at roughly half the cost and is now the default for paid ChatGPT (Plus, Team, Enterprise). Luna is small, fast and cheap. All three run a ~1.05M-token context with 128K max output, across ChatGPT, Codex and the API; the gpt-5.6 alias routes to Sol.

API pricing per 1M tokens:

OpenAI calls 5.6 its strongest cybersecurity model yet at far fewer tokens; Altman said Sol is about 54% more token-efficient on coding. Under the Preparedness Framework all three rate "High" in cybersecurity and bio/chem, but none reach "Critical."

Two side launches landed alongside the models. ChatGPT Work is an enterprise workplace companion for docs, sheets and decks. GPT-Live is a set of new voice models that listen and speak simultaneously. The efficiency angle is the one to watch: cheaper tokens make Terra the sensible default and widen OpenAI's margin. The bigger open question is regulatory. OpenAI just exited the govt-gated preview while telling Washington it should not become the norm, so watch whether the next frontier model gets the same review or skips it.