Its GPT-5.6 model family is designed to deliver more useful intelligence per dollar across models, inference and agent workflows. The company describes three tiers: Sol as the flagship, Terra as a lower-cost option and Luna as its fastest and most affordable model.
The practical claim is that capability and cost are being optimized together. OpenAI says GPT-5.6 Sol with maximum reasoning beats Claude Fable 5 on the Artificial Analysis Coding Agent Index at less than half the cost, while Terra performs like GPT-5.5 on intelligence benchmarks at half the price. Luna is priced 80 percent below Sol.
Those numbers matter for developers because model cost shapes which applications are economically viable. Cheaper capable models can make high-volume coding, customer support or agent tasks easier to justify.
The details are still based on OpenAI’s framing and selected benchmarks. Teams will need to test the models on their own workloads, where latency, accuracy, safety behavior and tool reliability may matter as much as headline price cuts.