RunMyLLM: A fit checker for local language models

Why it surfaced

It accounts for quantised weights, KV cache, context length, and estimated tokens per second rather than stopping at a vague hardware recommendation. The tool covers 91 devices, accepts arbitrary Hugging Face model IDs, and can provide the command needed to run a selected model.

RunMyLLM matches open-weight LLMs to your GPU or Apple Silicon machine and estimates whether they will run.