Cost-check any AI API in 30 seconds. ? How it works Enter your monthly usage and we rank every provider from cheapest to most expensive. For text: fill tokens-in / out / requests for accuracy, or just total tokens (we assume a 70/30 input/output split). Presets autofill realistic values. Prices are list rates — no Batch (-50%) or caching (-90%) discounts. An asterisk (*) flags providers with pricing caveats.

How we compute it

• We multiply your monthly usage by each provider's published API rate and sort the list from cheapest to most expensive.
• For text, the formula is (tokens_in × in_rate + tokens_out × out_rate) × requests / 1M. If you only enter total tokens, we assume a 70% input / 30% output split.
• Prices are list rates verified against each provider's official pricing page on 2026-10-26. They don't include Batch API discounts (-50%) or prompt caching (-90% on cached input).
• An asterisk (*) next to a provider means the rate has caveats — for example, GPT-5.6 Sol's $4/$20 is a promo through at least Nov 21 2026, DeepSeek bills peak vs off-peak (catalog shows off-peak), Jev (TypeSafe) bills input only at $0.042/1M with free output because it is a decision model not a chat LLM, and GPT Image 2's $4/$15 figure is the Batch rate (the row uses the $8/$30 standard list). Read the italic note below the row.
• Need a SaaS subscription instead of an API? Try the Find my AI wizard — it ranks plug-and-play tools by your monthly budget.