How much does GPT-5.6 cost?

As of July 2026, GPT-5.6 API pricing comes in three tiers: Sol costs $5 per million input tokens and $30 per million output tokens, Terra costs $2.50 and $15, and Luna costs $1 and $6. A million tokens is roughly 750,000 words, so a typical request costs fractions of a cent on Luna and a few cents on Sol.

Two levers cut the real bill well below list price. Prompt caching gives a 90% discount on repeated input — cached reads cost $0.50 per million on Sol, $0.25 on Terra, and $0.10 on Luna (cache writes bill at 1.25× the normal input rate). And OpenAI's Batch and Flex modes take 50% off standard rates for work that can run asynchronously.

Stacked together, a production workload with a stable system prompt often lands at roughly half of what the sticker price suggests.

In ChatGPT, GPT-5.6 Sol is included from the Plus plan rather than billed per token. For API budgeting, the practical move is to price your actual prompt sizes and volumes rather than reason from list rates alone — a realistic estimate needs your tokens per request, requests per day, and expected cache-hit rate.