Show HN: LLM Sorter – Python package to sort lists of items using LLM calls (github.com)

A new Python library called **LLM Sorter** has been launched, enabling users to sort lists of items based on semantic meanings using large language models (LLMs) as comparators. This innovative tool allows sorting by various criteria expressible in natural language, such as tone, urgency, and complexity. By leveraging an OpenRouter API key for LLM integration, users can sort lists with simple code snippets. For example, you can sort strings by reading complexity or prioritize support tickets based on urgency, facilitating tasks that require nuanced evaluation beyond numerical data. The significance of LLM Sorter for the AI/ML community lies in its ability to perform semantic sorting without needing extensive training data, making it a practical solution for rapid prototyping and internal tools. By employing an implementation of merge sort that utilizes LLM-driven comparisons, the library provides a way to tackle semantic properties difficult to quantify. However, there are limitations, such as potential cost and latency with larger lists, non-deterministic outcomes, and biases depending on phrasing and context. This tool is ideal for small to medium-sized lists where approximate "human-like" rankings are sufficient but may not be suitable for situations requiring strict sorting determinism or reproducibility.