Overview
Liquid Inference is a price-oriented service for AI inference requests. Instead of presenting a single fixed provider offer, it has providers compete on the price of every prompt and selects an offer according to the rules attached to that request.
The core idea is to connect each AI request with the lowest-priced provider offer that still meets the requester’s stated requirements.
How Request Pricing Works
Competition takes place for each individual AI request rather than being described as a one-time comparison. Multiple providers can compete on price, while the requester’s rules remain the qualifying condition for which offers are acceptable.
The amount paid is the lowest qualifying offer. This combines price competition with request-level constraints, so a cheaper offer does not qualify unless it also satisfies the applicable rules.
Primary Use Case
Liquid Inference is intended for obtaining AI inference through competitive provider pricing. Its focus is the economic selection of a provider for each prompt: providers submit competing prices, qualifying offers are determined by the request’s rules, and the lowest acceptable offer establishes what the requester pays.