Show HN: InferBench – Benchmark local LLM engines with one click
Category: devtools
Tags: llm-benchmark, electron-app, machine-learning
Score: 7.5/10 (Innovation: 7, Technical: 8, Documentation: 8, Utility: 7)
InferBench is a cross-platform desktop application that benchmarks local LLM inference engines with one click, automating model download, engine setup, and performance measurement. It stands out by combining auto-hardware detection, optimizer-driven quantization selection, and support for multiple engines (llama.cpp, Ollama, vLLM) in a user-friendly Electron+FastAPI stack. The project solves a practical problem for developers evaluating local LLM performance without manual command-line work.
Target audience: AI/ML engineers and developers who run local LLMs and need to compare inference engine performance
Repository: https://github.com/JoniMartin27/inferbench · Python · MIT
View on Hacker News