Free up spaceFree

Local AI models

Find the models filling your disk, including the ones you have twice.

Sweep lists the models you downloaded with Ollama, LM Studio and Hugging Face tools, with their size and date. It points out the same model kept by two apps, and removed models go to the Trash.

Junk Files listing local AI models from Ollama, LM Studio and Hugging Face, with a model kept by two apps pointed out
What it lists

Every model, measured properly

  • Ollama models, measured by the files only that model uses
  • Ollama model files no installed model uses
  • LM Studio models, with publisher and format
  • Hugging Face models and datasets in the hub cache
  • Hugging Face's download cache
  • Models from GPT4All, PyTorch Hub and Whisper
  • The same model file kept by both Ollama and LM Studio
How it works

Found where your apps really keep them

  1. 1

    Finds each app's folder

    Sweep follows each app's own settings and environment variables, so models in a custom folder are found too.

  2. 2

    Spots the same model twice

    Models kept by both Ollama and LM Studio are compared by size, then by a sampled hash, and marked as the same model.

  3. 3

    Leaves the choice to you

    No model is ticked for you, because big ones take a while to download again. Removed models go to the Trash.

Settings › Integrations

See where your models live

Integrations shows each app's model folder, how much it holds and whether Sweep found it. Point Sweep at another folder, or open it in Finder.

  • Ollama, LM Studio and Hugging Face
  • Change… and Show in Finder for each
  • Docker's status in the same place
Integrations: where Ollama, LM Studio and Hugging Face keep their models, and Docker's status

AI model questions

Something else? Ask us.

Clean, protect and understand your Mac.

Finds junk, big files and leftovers, checks for malware and risky settings, keeps apps updated, and shows what your Mac does when you're not looking. Free and open source.