How to Deploy DeepSeek Locally: Step‑by‑Step Guide for Offline AI

Local vs Online Versions – Pros and Cons

Advantages of the local version:

Data privacy is guaranteed because all processing happens on the device.

Fast response without network dependency.

Customizable parameters to match hardware and specific use cases.

Advantages of the online version:

Easy to use through a browser, no installation required.

No hardware upgrades needed; works on any device with internet.

Disadvantages of the local version:

Requires a GPU (GTX 1060 6 GB or higher) and sufficient RAM.

Higher technical barrier; environment setup and model tuning can be challenging.

Disadvantages of the online version:

Performance depends on network stability.

Data is sent to the cloud, posing privacy risks.

Hardware Requirements

Minimum GPU: GTX 1060 6 GB; recommended RTX 3060 or higher. Memory: at least 8 GB (16 GB recommended). Free disk space: 20 GB on an NVMe SSD.

Installation Tutorial

Download the Ollama client from https://ollama.com (Ollama is an open‑source tool for local LLM deployment).

Open a terminal (Win+R → cmd) and verify Ollama installation with ollama -v.

Search for deepseek‑r1 in Ollama, choose the model size that matches your hardware (e.g., ollama run deepseek-r1:1.5b for the 1.5 B parameter model), and copy the command.

Run the copied command to download the model. If the download stalls, press Ctrl+C and re‑run the command to resume.

After the model finishes downloading, you can start using DeepSeek locally.

For a browser‑based UI, install the “Pageassist” extension from the Chrome/Edge store, enable developer mode, drag the extension file into the extensions page, and add it.

Open the Pageassist UI, select the downloaded DeepSeek model, and configure language and speech settings to Chinese.

Note: To remove a model, replace run with rm in the command.