Ollama & Local AI Model Cleaner for Mac
Running Llama 3, DeepSeek, Mistral, or Qwen locally? Each quantized GGUF weights file takes 4–50 GB. CleanYourMac Space visualizes all model weights across Ollama and LM Studio so you can remove unused ones in seconds.
Where local AI models hide on macOS
When you run ollama run or download models in LM Studio, the neural network blobs are stored in hidden directories across your system:
Raw SHA-256 binary blobs for Llama, DeepSeek, Gemma, and embedding models.
Downloaded GGUFs in ~/.cache/lm-studio or ~/Library/Application Support/LM-Studio.
Transformers snapshot caches in ~/.cache/huggingface/hub.
How CleanYourMac Space Simplifies AI Storage
Unified AI Model Directory Map
Aggregates all AI models across Ollama, LM Studio, Jan, and HuggingFace into a single clear view with exact byte measurements.
Trash-First Safety
Moves selected model weights to Trash first. If you decide you still need an old model checkpoint, restore it immediately without re-downloading gigabytes.
No Command-Line Parsing Needed
No need to manually decode cryptographic SHA-256 blob names in ~/.ollama. The treemap visualizes exactly what is on your disk.
100% On-Device Privacy
Zero network requests. Your model names, prompts, and local datasets remain strictly confidential on your Mac.
Step-by-step guides
Prefer to do it by hand? These guides show the exact Terminal commands.
Clear 50+ GB of stale AI models now.
Download CleanYourMac Space free to inspect your model caches, or upgrade to Pro for $19.99 lifetime.