OpenAI releases GPT-5.6: flagship Sol, Terra, and budget-friendly Luna

OpenAI has moved its GPT-5.6 model family out of limited preview testing and into general availability. The lineup includes three models: the flagship GPT-5.6 Sol, GPT-5.6 Terra, which is balanced for everyday work, and GPT-5.6 Luna, the most cost-efficient of the three.

What can the flagship Sol do?

According to OpenAI, GPT-5.6 Sol delivers best-in-class results in coding, knowledge work, cybersecurity, and science, outperforming both earlier OpenAI versions and competing frontier models. It also uses fewer tokens and costs less per task, so users get more work done on the same budget.

• The family consists of three models: Sol (flagship), Terra (balanced), and Luna (cost-efficient)
• Sol leads in coding, knowledge work, cybersecurity, and science
• A new ultra mode coordinates multiple agents across parallel workstreams
• The family previously went through limited preview testing under the codename Sol

Why ultra mode?

OpenAI introduced ultra, a maximum-capability mode in which several agents work in parallel on different parts of a single task. This should help complex projects finish faster than they would with a single agent working sequentially.

The company also points to more confident computer use and sharper design sense: OpenAI positions GPT-5.6 Sol as its most polished collaborator to date, one that can check, refine, and deliver output ready to use directly, without additional polishing.

What it means

GPT-5.6 continues the trend of OpenAI's recent releases: quality gains now come with a lower cost per request, not just expanded capabilities. The arrival of a separate low-cost Luna variant and the parallel ultra mode shows the company is extending its lineup in both directions at once: down, for high-volume tasks, and up, for complex multistep projects that require coordinating multiple agents.