Compresr

Compresr is an API designed for context compression specifically tailored for large language models (LLMs). The service allows users to input a long context along with a query, and it returns a significantly shorter context that retains the essential answer-bearing tokens while discarding the rest. This functionality is aimed at reducing both token costs and latency, with the promise that at light compression levels, it can match or even exceed the accuracy of full-context responses. Compresr offers a Python SDK, TypeScript SDK, and a hosted HTTP API, making it accessible for various developme…