llmfit

llmfit compares language models with your CPU, RAM and GPU capacity.
Browse recommendations in the terminal, produce JSON reports, download
models and connect to running inference servers such as Ollama.

Connect hardware access before starting:

sudo snap connect llmfit:hardware-observe

sudo snap connect llmfit:system-observe

llmfit

The web dashboard is available with llmfit serve at http://localhost:8787.
GPU detection uses kernel and PCI information; NVIDIA memory may be
estimated. Use --memory to override it. Host inference executables and
hidden model directories are not accessible from this strict snap.