Liquid Inference

Liquid Inference is an auction exchange for LLM inference. Providers post standing offers. The exchange ranks the eligible offers for each request, sets a price cap before the first token, and charges for the tokens used.

This plugin adds Liquid Inference as a model provider in Dify.

Setup

1. Create a workspace on Liquid Inference, add a balance, and create a key.
2. In Dify, open Settings > Model Provider and install Liquid Inference.
3. Paste the key in API Key. Leave API Base URL empty to use .

Models

is in the list. It lets the exchange choose the model for each request.

To use another model, choose Add Model and enter its name exactly as returns it. Set the context size and the features of that model.

Cost

The exchange sets a cap on each request. It sends the cap in the response header. To limit total spend, set a spending limit on the key.

Payments

The plugin makes no payments and moves no funds. It sends chat requests only. Liquid Inference bills your Liquid workspace balance, which you manage on the Liquid Inference site.

Source

https://github.com/architect-xyz/liquid-dify-plugin