OpenAI announced price and performance updates for the GPT-5.6 family aimed at lowering costs for high-volume use and speeding up time-sensitive workloads. The company said these changes stem from efficiency improvements across models, inference systems and the agentic harness that connects models to tools and context.
Price reductions and Fast mode
Starting July 30, 2026, OpenAI said GPT-5.6 Luna will be 80% less expensive and GPT-5.6 Terra 20% less expensive. Luna is positioned as the fastest, most affordable option for large-scale tasks, while Terra remains the balanced model for everyday work. The company also introduced Fast mode for GPT-5.6 Sol, which OpenAI says delivers up to 2.5× faster responses than Standard processing at twice the price; Fast mode replaces the previous Priority Processing option and is backward compatible with priority-tagged requests.
Efficiency gains and measured results
OpenAI attributed the savings to optimizations at multiple levels. The firm reported that Sol helped rewrite and optimize production kernels, running hundreds of experiments to improve token generation; those kernel changes reduced end-to-end serving cost by about 20% and raised token-generation efficiency by more than 15%, according to OpenAI.
OpenAI highlighted performance comparisons: Luna reportedly delivers intelligence comparable to frontier-class models from a year earlier at roughly “6 cents on the dollar per task” and at nearly nine times the speed. On a benchmark called Agents’ Last Exam, OpenAI said Luna outperforms Fable 5 at an estimated cost per task nearly 99% lower. A customer quote from Michele Catasta, President & Head of AI at Replit, said Luna is “the closest we’ve come to intelligence too cheap to meter.”
Availability and specific pricing
OpenAI said Terra and Luna remain available in ChatGPT Work, Codex and the OpenAI API. In ChatGPT Work and Codex, Free and Go users can access Terra; Plus, Pro, Business and Enterprise users can choose Terra and Luna. OpenAI published API token prices effective July 30: Terra at $2 per million input tokens and $12 per million output tokens; Luna at $0.20 per million input tokens and $1.20 per million output tokens. Sol pricing was left unchanged. The company said ChatGPT and Codex subscription prices and quota budgets remain unchanged, while Terra and Luna usage will consume fewer credits, and that pricing changes will begin rolling out in AWS later the same day.
Original source: OpenAI News