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

OpenAI announced the transition of the GPT-5.6 family of models from limited preview mode to general availability. The lineup includes three models: the flagship GPT-5.6 Sol, GPT-5.6 Terra balanced for everyday work, and the most economical GPT-5.6 Luna.

What the flagship Sol can do

According to the company, GPT-5.6 Sol shows best-in-class results in programming, knowledge work, cybersecurity, and science — surpassing both previous OpenAI versions and competing cutting-edge models. At the same time, the model uses fewer tokens and is cheaper in terms of cost per unit, giving users more work done for the same budget.

• The family consists of three models: Sol (flagship), Terra (balanced), and Luna (economical)
• Sol leads in programming, knowledge work, cybersecurity, and science
• A new ultra mode has been introduced — coordinating multiple agents in parallel workstreams
• The family previously underwent limited preview testing under the codename Sol

Why the ultra mode is needed

OpenAI introduced ultra — a mode of maximum capabilities in which multiple agents work in parallel on different parts of one task. This should accelerate the completion of complex projects compared to sequential work by a single agent.

The company also notes more confident computer use by the model and more precise design intuition: GPT-5.6 Sol is positioned as OpenAI's most refined partner today, capable of checking, refining, and delivering results ready for direct use without additional refinement.

What does this mean

GPT-5.6 continues the trend of OpenAI's recent releases: growth in model quality is now accompanied by reduced request costs, not only by increased capabilities. The appearance of a separate economical Luna option and a parallel ultra mode suggests that the company is simultaneously expanding the lineup downward — for mass-market tasks — and upward — for complex multi-step projects requiring coordination of multiple agents.