RunMyLLM is an online guide to running open-weight large language models on personal computers and other local hardware. It helps users identify which models their GPU or Apple Silicon device can run and recommends options for tasks such as coding, reasoning, vision, agents, and general chat. The catalogue covers 91 profiled devices and 79 models, with updates to newly released models each week.

Users can select a device or paste a Hugging Face model ID or link to assess whether a model will fit. RunMyLLM estimates quantised model-weight sizes, adds the KV cache required for the chosen context length, and provides estimated token speeds. Results group models by whether they run comfortably, fit tightly, or exceed the hardware’s capacity. Model pages also provide commands for running models with Ollama, llama.cpp, LM Studio, or MLX, with context settings included. The site further offers hardware recommendations for specific workloads, explanations of its sizing approach, and a comparison resource for considering local hardware costs alongside hosted AI services.