OpenAI GPT-5.6 is here: Sol, Terra and Luna now available

Update, July 11, 2026: This article originally described the limited preview. OpenAI made GPT-5.6 generally available on July 9, so the title and text have been corrected.

OpenAI has made GPT-5.6 generally available. The company’s most capable model to date comes in three tiers, Sol, Terra and Luna, and has been available since July 9, 2026 in ChatGPT, in Codex, through the API and in GitHub Copilot. The rollout ran globally over about 24 hours. Sol is the flagship, Terra the balanced option for everyday work, and Luna the fast and affordable tier.

Where you can use GPT-5.6 now

In ChatGPT, Sol is available to users on the Plus plan and above, including Pro, Business and Enterprise. In Codex, Plus and higher can choose between Sol, Terra and Luna and set a reasoning effort per model, and the new ultra mode is available there from Plus as well. On top of that come API access and the integration into GitHub Copilot. If you want to know how ChatGPT, Gemini, and Claude perform in everyday use, you can find a direct comparison here.

What sets Sol, Terra and Luna apart

The naming is new. The number stands for the generation, while Sol, Terra and Luna denote three durable capability tiers. Terra matches the performance of its predecessor GPT-5.5, according to OpenAI, while costing only half as much. New features include a max level for especially long reasoning and an ultra mode that distributes subtasks across multiple sub-agents. Prices per million tokens are $5 for input and $30 for output on Sol, $2.50 and $15 on Terra, and $1 and $6 on Luna.

Why the launch was initially throttled

Before the broad release, GPT-5.6 ran in a limited preview for only a small group of selected partners. The background was its significantly increased capabilities in cybersecurity. OpenAI had shown the U.S. government the model’s capabilities in advance and deliberately kept the group small at first. According to OpenAI, GPT-5.6 is particularly good at identifying vulnerabilities and fixing bugs, which helps defenders but poses risks in the wrong hands. OpenAI stresses that such a government-led access process is not meant to become permanent. With general availability now here, that preliminary stage is complete.

Speed boost through specialized hardware

One detail is likely to interest tech enthusiasts. In July, OpenAI is also bringing GPT-5.6 Sol to hardware from chip specialist Cerebras, at up to 750 tokens per second. That is many times the usual speed on traditional graphics cards and shows that the industry is increasingly looking for alternatives beyond the well-known GPU market leaders. This access starts initially only for select customers while Cerebras expands capacity.

The bottom line is that GPT-5.6 is a major leap, and unlike at the preview start, most people can now try it out for themselves right away. In ChatGPT, in Codex and through the API, Sol is ready, with Terra and Luna providing the cheaper tiers.