Contribute a benchmark

Help everyone get accurate numbers. Run llama-bench or ollama run --verboseon your machine, then tell us what you ran and paste the output — we'll extract the tokens/sec and add it to the measured data. Anonymous.

Your computer

Model

Quant

Context

Benchmark output

Protected by a small proof-of-work check — no third-party CAPTCHA, no tracking.