llm-switchboard

Route prompts to local large language models instantly to cut costs and reduce latency with smart, zero-overhead classification under 1 millisecond.

GitHub

Official source

Start locally with the CLI guide