Step-by-Step Guide to Deploying DeepSeek Locally with Cherry Studio

Recent popularity of DeepSeek has led many to seek a smooth deployment experience, but server overload and hardware limits often hinder local runs of 7B/14B models.

By leveraging SiliconFlow and Cherry Studio, you can run the full‑blood DeepSeek version on any internet‑connected computer without demanding specifications.

Online Experience

1. Preparation

Register an account on the SiliconFlow website (https://siliconflow.cn/zh-cn/) by clicking the top guide or Login button. New users receive 20 million tokens for testing.

2. Model Selection

Choose DeepSeek‑R1 (the 671B full‑blood version) or other models such as DeepSeek‑V3 or deepseek‑ai/Janus‑Pro‑7B for image generation.

Switch models directly in the interface to try chat and image generation features.

Private Deployment

1. Install Cherry Studio

Download the client from https://cherry-ai.com/ and install it; it supports multi‑provider AI integration.

2. Obtain API Key

In the SiliconFlow account, create a new API secret key and copy it.

3. Environment Configuration

In Cherry Studio, go to Settings → SiliconFlow, paste the API key, select the desired model (e.g., deepseek‑ai/DeepSeek‑R1), and save.

4. Verification Test

Start a new conversation to confirm interaction works. Cherry Studio also provides preset roles (e.g., “Test Engineer”) to streamline prompt engineering for tasks like test‑case generation.

Conclusion

Following these steps, you can set up a full‑feature AI assistant within minutes, ensuring local storage of conversation data and privacy protection, while leveraging powerful DeepSeek models without high‑end hardware.