Architect Launches Liquid Inference Auction Platform

Architect Financial Technologies announced in a press release the launch of Liquid Inference, a live auction platform for large language model inference. Providers submit competing offers, and the platform awards each request to the lowest priced offer that meets the buyer's requirements.

Buyers can set limits for cost, time to first token, throughput, approved regions and data retention. They can also restrict requests to specified providers or models. The platform publishes live order books and quotes, locks a maximum price before generation begins and charges for metered use.

Liquid Inference provides a chat interface and a standard model API. It supports OpenAI and Anthropic compatible clients, including Claude Code, Codex, Cursor and Cline. Each completed request generates a receipt showing the selected provider, price cap, final charge and competing offers available when the request was awarded.