Manage Gigabytes of Local AI Models on macOS
Running 7B, 13B, and 70B models eats high-speed Mac SSD space fast. DiskWarren automatically uncovers Ollama blobs, LM Studio GGUFs, Hugging Face snapshots, and ComfyUI checkpoints.
Ollama Models (~/.ollama/models)
Ollama stores model weights as SHA-256 binary blobs without recognizable names. DiskWarren parses the manifest index to show exact model names, sizes, and last modified dates.
LM Studio & GGUF Weight Files
Locate quantized GGUF weights, verify their binary headers, and identify duplicate quantizations (e.g. Q4_K_M vs Q8_0) consuming redundant space.
Hugging Face Hub Snapshots
Hugging Face caches full model revisions in ~/.cache/huggingface/hub. Identify unused transformer weights and safely reclaim space.
ComfyUI Checkpoints & LoRAs
Diffusion models and LoRA adapters take up massive space. DiskWarren indexes your models folder to give you complete visibility.