Local LLM hardware calculator
Pick a model and your machine and see whether it fits, which quantization to use, and roughly how fast it will run. Sizes are verified against the Ollama library where published and derived from the model math everywhere else. Your answers stay in the browser.
Every model in the catalog, at Q4_K_M with the context you pick. Click a row to open the full check.
Weight sizes are verified against the Ollama library at update time where a Q4_K_M GGUF exists, and derived from parameters times bits per weight everywhere else. KV cache and overhead are estimates with the honest margin built in. Model data updated: September 2026.