LM Studio is one of the most user-friendly ways to run local AI models on Mac. That user-friendliness is also why it is one of the most common sources of unexpected storage usage — downloading a model is three clicks, and LM Studio makes no particular effort to remind you how much you have already downloaded.
How the storage accumulates
LM Studio stores model files in a directory you can change from the settings, but most users leave it at the default. Each model is a GGUF file — a format designed for efficient local inference. Sizes range from around 2 GB for small models to over 40 GB for large ones.
The accumulation pattern is predictable:
- You download a model to test it
- You download a newer version of the same model when it is released
- You download a different quantization of the same model to compare quality versus speed
- You forget the first version is still there
Three versions of one model and you have potentially committed 15+ GB to a single model family.
Want to skip the hidden-folder hunt?
LLM Cleaner scans your Mac for local AI models, caches, indexes, and project memory — then shows what you can review, reveal, export, or safely move to Trash.
The duplicate problem between LM Studio and Ollama
LM Studio and Ollama serve similar purposes — they both run large language models locally on your Mac. Many developers use both. What many do not realise is that they may have downloaded the same underlying model through both tools.
Mistral 7B in Ollama and Mistral 7B in LM Studio are two separate files. They do not share storage. They do not know about each other. You can have the same model twice on your machine, taking double the space, with no warning from either application.
Spotting these duplicates manually means checking the contents of both tools' model directories, comparing sizes and names, and making a judgment call about which version to keep. That is a task most developers never actually do.
What to do before deleting
The safest approach before removing any LM Studio model is to see the full picture: what is present, what size each model is, when it was last used, and whether the same model also exists in another tool's storage.
Removing a model from LM Studio without checking whether Ollama has it too (or vice versa) means you might be keeping the duplicate you did not intend to keep.
Local AI storage is messy. Cleanup should not be.
Find Ollama, LM Studio, Hugging Face, Cursor, Windsurf, Claude Code, Whisper, and loose model files in one place.
Custom model directories
LM Studio lets you change the directory where models are stored — useful for keeping large files on an external drive. If you have changed this setting at any point, you may have models in two locations: the original default directory and the new custom one, with no clear indicator of which is current.
Not sure what is safe to delete?
LLM Cleaner separates models, rebuildable caches, and project memory so you do not treat everything like junk.