How To Get The Lowest Cost Llm Inference On Openrouter
Source
openrouter.ai
Author
OpenRouter
Date
Context
OpenRouter's cost guide distinguishes choosing the cheapest provider from refusing an expensive request. Adding the :floor suffix prioritizes providers by price for a chosen model. Setting max_price instead excludes providers whose input or output rates exceed the specified ceilings; if none qualify, the request fails.
That distinction lets an application choose a tradeoff explicitly. It can prioritize throughput among providers below a price ceiling, rather than always selecting the lowest price. These settings govern provider rates per million tokenThe chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.Full definition →. A lower rate still leaves total spending dependent on how much the application generates and consumes.
Terms in this piece · Glossary
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
quantization — Shrinking a model by storing its numbers less precisely — like rounding — so it runs faster and fits on smaller hardware, at a small quality cost.
