# Ollama on NAS: Storage, GPU, Docker and Remote Compute Guide

> For most capable home and small-office setups, keep source data and backups on the NAS, put frequently used Ollama model weights on NVMe close to the inference host, and use the NAS over 2.5GbE or 10GbE for datasets and archives. Run Ollama directly on the NAS only when the NAS has enough memory, supported acceleration and cooling for the model size you intend to use.

Canonical: https://cloudzat.com/ollama-on-nas/

Use the canonical page for live hardware availability and current deployment guidance.

## Related Cloudzat guidance
- [Open WebUI on NAS](https://cloudzat.com/openwebui-on-nas/)
- [AnythingLLM on NAS](https://cloudzat.com/anythingllm-on-nas/)
- [n8n on NAS with Local AI](https://cloudzat.com/n8n-on-nas-local-ai/)
- [Best AI NAS Software](https://cloudzat.com/ai-nas-software/)