Unterstütze llm-bench.io

I got tired of guessing. Running local models on my Radeon RX 7900 XTX, I burned way too many evenings benchmarking models to find the right one for each use case — so I built the tool I wished existed.

llm-bench.io is a community-driven benchmark for local LLMs: real speed, memory, and quality numbers across real hardware, so you can pick the right model in minutes instead of days.

It's free and always will be — but it isn't free to run. Your support goes straight into keeping the lights on:

🖥️ Hosting & database for the site and all the benchmark data

🤖 AI quality-scoring of every submitted benchmark

🛠️ New features — more models, more hardware, better comparisons

If it saved you some time on your local-LLM journey, a coffee helps me keep it alive and growing. Thank you! ☕