Metrillm

MetriLLM is an open-source benchmarking toolkit for local LLMs (Large Language Models) that lets developers quickly evaluate the speed, quality, and hardware fit of models on their own machines. With a CLI, MCP server, and integrations for IDEs like Claude Code and Cursor, MetriLLM delivers comprehensive performance and quality metrics, generates global scores, and allows effortless sharing via a public leaderboard. It's perfect for developers, AI engineers, and hobbyists looking to compare, optimize, or select the best LLM for their hardware and use case.