OpenAI Cuts Luna Pricing by 80%: Volume Workloads Recalculated

On 30 July, OpenAI cut API prices for two models in the GPT-5.6 family, which launched on 9 July. Luna fell by 80%, Terra by 20%.

Which tiers got cheaper — and why not the flagship

A million input tokens on Luna now costs $0.20 instead of $1, and a million output tokens $1.20 instead of $6. Terra's input dropped from $2.50 to $2, output from $15 to $12. The flagship Sol stayed at $5 per million input tokens, and gained a priority processing mode for requests.

Why rejected business cases are worth reopening

High-volume traffic usually runs on mid- and low-tier models: sorting inbound email, drafting replies, tagging documents. That is exactly where the price fell fivefold on input and fivefold on output. Features shelved a year ago because each call cost too much now deserve a second look — at the same request volume, the monthly bill changes by a multiple. The recalculation takes an evening: take the average request size in tokens and the number of calls per month, then multiply by the new rate. A budget calculator gives a quick sense of the order of magnitude.

What remains unclear

Whether the new rates apply through Russian resellers, what the payment terms and timelines from Russia look like, whether context window limits or rate limits have changed. Until those questions are answered, any calculation is an estimate rather than a budget.