OpenRouter's routing shortcuts change which provider serves a chosen model. The :nitro suffix orders providers by throughput, meaning 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 → generated per second, while :floor orders them by price. The same preferences can be set explicitly through the API. The announcement also lists latency as a separate sorting option, so choosing Nitro should be understood as prioritizing generation speed rather than necessarily the shortest wait before a response begins. Developers can express that preference without changing the underlying model.
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.
