Llmspeed: test local LLM speed before buying hardware

Llmspeed is a simulator that lets developers test how fast a local large language model will run on different hardware before making a purchase. Users select a GPU, Mac configuration, or other device and a specific model, and the tool shows simulated tokens per second and time-to-first-token metrics based on verified benchmark data.

The tool addresses a real pain point: buying expensive hardware without knowing whether it will deliver acceptable performance for the LLM workload you need to run. By letting people experiment with hardware and model combinations first, it aims to help reduce wasted spending and support faster decision-making.