Nic Schlueter
Is there any website like Geekbench that shows tokens / second an LLM can generate, by machine, model, context, quantization etc?
Mar 25, 2025 at 11:50amWeb
Show threadStephan HeuelMar 25, 2025
@schlu @martinhoeller
Good question. I have seen https://kamilstanuch.github.io/LLM-token-generation-simulator, but not sure about accuracy.
LLM Token Generation Speed Simulator & Benchmark | Compare Local LLM Performance

Compare LLM token generation speeds across devices and models. Benchmark your hardware for local LLM inference and find the best setup for your needs.

Show threadNic SchlueterMar 25, 2025
@ping13 @martinhoeller Thanks for that link. It looks good, but a bit thin on detail, but I do appreciate it.