llmfit: Automatic LLM model matching for your PC specifications

Automatically detects RAM, CPU, and GPU specifications to select models from hundreds of LLMs that run well on your computer. It scores models based on four criteria—memory compatibility, estimated speed, quality, and context length—to help you assess actual feasibility.

Rather than relying on simple estimates, it calculates speed based on memory bandwidth models and actual community measurement data. It is designed to provide the input values and verification commands used for each estimate, allowing users to verify them directly. It supports various local runtimes such as Ollama, llama.cpp, and MLX, as well as multi-GPU and MoE architectures.

It offers both interactive TUI and CLI modes. The TUI allows you to view hardware specifications and model rankings at a glance and run benchmarks, while the CLI supports JSON output for script or agent integration. You can handle the entire process from model download to serving locally within the terminal.

It is suitable for developers who want to set up a local LLM environment but are unsure which model to choose. It can be used when you need realistic model recommendations considering hardware constraints, or when you want to automate model compatibility checks in CI/CD pipelines.