GPT-5.6 Luna is the fastest and most affordable variant in OpenAI's GPT-5.6 lineup, sitting alongside the flagship GPT-5.6 Sol and the balanced GPT-5.6 Terra as part of a three-tier family rolled out together on Vercel AI Gateway on July 9, 2026 in a limited preview. OpenAI built Luna for high-volume, cost-sensitive workloads, framing it as a way to run tool-driven, multi-step enterprise flows at scale without sacrificing responsiveness. Official documentation lives on the OpenAI API docs under the Models section, where developers can find guidance on text and code generation, structured outputs, reasoning, images and vision, and tool use that apply to the Luna tier as well.

Practical fit for Luna centers on throughput and price rather than frontier reasoning: OpenAI's July 30, 2026 price-performance announcement cut Luna's pricing by 80% and explicitly highlighted its ability to use tools and complete multi-step workflows, aiming for frontier-level intelligence at a fraction of the cost of similarly capable models. Through Vercel AI Gateway, it is exposed via a unified API with usage and cost tracking, retries, failover, Zero Data Retention support, and per-key or team controls, and can be routed to from other gateway models with no code changes by setting the slug or adjusting gateway rules. Independent downstream corroboration from products like TypingMind reinforces Luna's positioning as the family's quickest, most economical option for everyday, high-volume requests.