Deploy a Private DeepSeek Large‑Model on JD Cloud with Ollama

DeepSeek has become a hot topic in the tech community for its powerful big‑data processing capabilities, but the official website often responds slowly, prompting many users to seek a private deployment.

Why Deploy a Private Large‑Data Model?

Deploying a private DeepSeek model offers several unique advantages:

Free usage after the initial setup, eliminating ongoing cloud service fees.

Data privacy protection by keeping all data locally.

No network dependency, allowing use anytime, anywhere.

Flexible customization to meet specific needs.

Performance and efficiency gains by leveraging local hardware.

No additional usage limits, enabling unrestricted experimentation.

DeepSeek Model Versions

DeepSeek provides two main families of models:

Full version – the complete model with massive parameters (e.g., DeepSeek‑R1 full version has 6710 billion parameters) requiring high‑end hardware such as 1 TB RAM and dual H100 80G GPUs.

Distilled version – smaller models created via knowledge distillation, ranging from 1.5 B to 70 B parameters, offering a better cost‑performance trade‑off.

DeepSeek‑R1‑Distill‑Qwen‑1.5B

DeepSeek‑R1‑Distill‑Qwen‑7B

DeepSeek‑R1‑Distill‑Qwen‑14B

DeepSeek‑R1‑Distill‑Qwen‑32B

DeepSeek‑R1‑Distill‑Llama‑8B

DeepSeek‑R1‑Distill‑Llama‑70B

Purchase and Log In to JD Cloud Computer

Visit the JD Cloud website, use the JD app to scan the QR code, and purchase a cloud computer (e.g., 4‑core 8 GB for ¥9.9 × 7 days). After ordering, follow the provided instructions to log in.

Set Up Ollama Service

Ollama is an open‑source LLM service tool that simplifies local model deployment.

Download the Ollama installer for Windows from the official website.

Run the installer and follow the prompts.

Verify installation by checking the task‑bar icon and running ollama in a command window.

Configure Ollama to start automatically: open services.msc, locate the Ollama service, set its startup type to Automatic , and confirm.

Run DeepSeek Model

After Ollama is running, pull and run the desired DeepSeek model version from the Ollama library.

Model Version

Required Space

Deploy Command

Configuration Note

1.5b

1.1 GB

ollama run deepseek-r1:1.5b

Suitable for 4 GB RAM computers, lightweight tasks.

7b

4.7 GB

ollama run deepseek-r1:7b

Fits 16 GB RAM machines, moderate tasks.

8b

4.9 GB

ollama run deepseek-r1:8b

Same as 7b, suitable for medium complexity.

14b

9 GB

ollama run deepseek-r1:14b

Higher complexity, needs more memory.

32b

20 GB

ollama run deepseek-r1:32b

High‑accuracy tasks on larger machines.

70b

43 GB

ollama run deepseek-r1:70b

Very high complexity, large memory.

671b

404 GB

ollama run deepseek-r1:671b

Extreme scale for massive knowledge tasks.

Configure Local Knowledge Base with CherryStudio

Download and install CherryStudio from its official site.

Add the DeepSeek‑R1 model using an API key.

Upload local documents (e.g., a novel) to build a knowledge base.

Select the knowledge base in the chat window to ask content‑specific questions.

Enable Network Access via Page Assist

Install the Page Assist browser extension.

Configure it to use the locally deployed DeepSeek model and select an embedding model.

Toggle the network switch to allow the model to perform web searches for up‑to‑date information.

Anything LLM for Local Knowledge Base and Network

Download and install Anything LLM.

Select Ollama as the model manager and configure embedding and vector database settings.

Upload documents to create a local knowledge base.

Adjust query settings for better relevance.

Enable optional network access for fallback searches.