How to Manage and Delete Local Ollama Models on Mac
Running local LLMs with Ollama is seamless, but model weights are huge. A single 70B quant is 40+ GB, and multiple models will rapidly fill even a 1TB SSD. Here is how Ollama organizes storage and how to prune it.
Where Ollama Stores Model Weights on macOS
By default on macOS, Ollama stores its manifests and binary layer blobs inside your home directory:
Because Ollama splits models into content-addressed SHA-256 blobs (similar to Docker), you cannot simply look inside this folder with Finder and see human-readable model names.
Method 1: Visual Management with DiskWarren
DiskWarren includes native Ollama intelligence. It parses your local manifests and translates cryptographic blob hashes back into human-readable model tags (e.g. llama3.3:70b-instruct-q4_K_M), showing exact sizes, parameter counts, and last accessed dates.
Method 2: Command Line CLI
You can also list and delete models via the Ollama CLI:
Inspect Local AI Storage Visually
Download DiskWarren for macOS • 100% Native & Private