AIAI > Local LLMs

AlexsJones/llmfit: See what LLMs will run on your laptop

github.com5 min readsaved Feb 16, 2026
Open original
TL;DR

llmfit is a Rust CLI tool that detects your hardware (RAM, CPU, GPU/VRAM) and tells you which of 48 popular open-source LLMs will actually run — scoring each as Perfect, Good, Marginal, or Too Tight. Features an interactive TUI and classic CLI mode. Available via brew or curl.

Why it matters

Instantly know which open-source LLMs you can run locally before wasting time downloading 40GB models that won't fit. Essential if you're experimenting with local inference.

Key points

  1. 1

    Detects RAM, CPU cores, NVIDIA/AMD/Intel Arc GPU VRAM via sysinfo and nvidia-smi/rocm-smi

  2. 2

    Database of 48 models: Llama, Mistral, Qwen, Gemma, Phi, DeepSeek, Cohere, and 9 more providers

  3. 3

    Memory computed using Q4_K_M quantization (0.5 bytes/param) — VRAM = primary constraint, system RAM = fallback

  4. 4

    Fit levels: Perfect (fits in VRAM), Good (GPU+CPU offload), Marginal (CPU-only), Too Tight

  5. 5

    Interactive TUI with search, filtering, and per-model detail view — or use --cli for table output in scripts