llmspeed is a browser-based simulator for exploring how local language-model generation might feel on different hardware. Users choose a hardware and model combination and can compare the pace of a simulated response before deciding what machine or model setup to investigate further. The site organizes its choices both by hardware and by model.
The project distinguishes measured community benchmark results from estimates derived from similar configurations or projected specifications. Its figures are presented for a short prompt on a single device, so they are not a guarantee for every workload or deployment. The simulator is open source, with its code and data linked on GitHub. It is a comparison aid for the experience of response speed, rather than a benchmark actually running on the visitor's computer.
