AI Model Atlas

469 AI models in one table

Provider prices converted to USD, context length, operating modes, and what no other site shows: a changelog of pricing updates. I pull data daily from OpenRouter's public API and exchange rates from public sources. Nothing here is manually copied or made up by a model.

Updated: 11 November 2026 Free models: 24 Providers: 61

How to read this table

Rates are shown per million tokens, because that's how every provider bills. A token isn't a word: in English it works out to roughly four characters per token, so an A4 page of text is about 600 tokens, and an average book fits in 180 thousand. The "1500-word article" column shows the same thing in USD, for a task someone would actually give the model: 1500 tokens of brief as input and 2200 tokens of finished text as output.

Context is the window the model can look at all at once. A million tokens sounds impressive, but it counts together with the response and the entire previous conversation, and the longer the window, the slower and more expensive it gets toward the end.

This table is only about pricing and parameters. Which model is smarter is decided by benchmarks, and you'll find those in the intelligence ranking. If you want to calculate the cost of your specific text, paste it into the token counter, because English tokenizes differently and the bill can surprise you.