llmsort

Score a list by any fuzzy attribute with an LLM judge: pairwise ratio questions fitted into consistent scores with error bars, at a known cost.

GitHub

Official source

Start locally with the CLI guide