GPT-5.6 Sol vs Terra vs Luna is the choice every OpenAI API user faced after the GPT-5.6 launch in summer 2026. OpenAI split GPT-5.6 into three models: Sol, the most capable; Terra, the balanced middle option; and Luna, the fastest and cheapest. Picking the right one can cut your AI bill dramatically without hurting quality.
This guide explains what each model is for, current pricing after OpenAI’s July 30 price cuts, and three essential tips to choose.
Table of Contents
GPT-5.6 Sol vs Terra vs Luna at a Glance
| Model | Role | Input / output per 1M tokens |
|---|---|---|
| GPT-5.6 Sol | Flagship, most capable | $5 / $30 |
| GPT-5.6 Terra | Balanced quality and cost | $2 / $12 (cut from $2.50 / $15) |
| GPT-5.6 Luna | Fastest and cheapest | $0.20 / $1.20 (cut from $1 / $6) |
Prices are the API list prices reported after OpenAI’s July 30, 2026 cuts. Some reports have mixed up Luna’s price with an older model, so confirm current figures on OpenAI’s pricing page before you budget.
What Each Model Is For
- Sol: hard reasoning, complex coding, long multi-step agent tasks, and work where a mistake is expensive.
- Terra: everyday business work such as drafting documents, customer support answers, summarising and analysing data.
- Luna: high-volume, simple jobs such as classifying emails, tagging, extracting fields from forms and short replies.
The GPT-5.6 Sol vs Terra vs Luna decision is less about which model is best and more about which is good enough for each task. Many businesses end up using two or three of them.
Pricing: GPT-5.6 Sol vs Terra vs Luna
Output tokens cost several times more than input tokens on all three models, so tasks that generate long answers cost more. Cached input, where the same instructions or documents are reused across requests, is billed at a large discount on all three. Sol also offers a fast mode at twice the standard price.
What It Costs in Practice
A rough example: a workflow that sends 10 million input tokens and receives 2 million output tokens a month would cost about $110 on Sol, $44 on Terra and $4.40 on Luna at the prices above. For a small business, the gap between Sol and Luna can be the difference between a noticeable bill and a rounding error.
3 Essential Tips to Choose
- Start with Terra, then test down and up: run your real prompts on Terra, then try the same prompts on Luna. If Luna’s answers are good enough, switch. Move to Sol only for tasks where Terra clearly falls short.
- Route by task: in tools such as n8n or Make, send simple steps (classification, extraction) to Luna and only the hard steps to Terra or Sol.
- Measure cost per result: track cost per resolved ticket or per finished document, not just the monthly total, and review it every quarter.
For more on building these workflows, see our guide to the best AI agents for business automation.
What About GPT-6?
OpenAI has since released GPT-6 models that use the same Sol and Luna naming. GPT-5.6 models remain available through the API, and if your workflows run well and cheaply on them, there is no rush to switch. Compare the newer options in our GPT-6 Sol vs Luna comparison.
The Bottom Line
In the GPT-5.6 Sol vs Terra vs Luna comparison, Terra is the best default for most small business tasks, Luna is the best value for high-volume simple work, and Sol is worth paying for only on your hardest problems. Test with your own prompts, route tasks to the cheapest model that does the job well, and check OpenAI’s pricing page regularly.