How to Deploy the Full-Feature DeepSeek LLM Locally and on Alibaba Cloud

Self-Deploy DeepSeek

Preparation: OS Ubuntu 20.04 LTS or CentOS 7, 16 GB RAM, 4‑CPU, 50 GB disk; optional network acceleration for faster downloads.

Basic deployment

# Install Docker and dependencies
sudo apt-get install -y docker.io docker-compose python3-pip git
sudo systemctl start docker

# Clone repository
git clone https://github.com/deepseek-ai/deepseek.git
cd deepseek

# Set environment variable
echo "API_KEY=your_api_key" > .env

docker-compose up -d

Quick start for beginners (Ollama)

ollama run deepseek-chat:8b  # 8B model requires ~5 GB storage

Enterprise usage (full features)

import deepseek
client = deepseek.Client(api_key="your_key")
response = client.chat(prompt="Your question", search_web=True)  # enable real‑time web search

Notes: keep the network stable when downloading the model (≈5 GB); use acceleration tools if needed; the enterprise API supports real‑time search and an extensible knowledge base.

Deploy DeepSeek on Alibaba Cloud

Alibaba Cloud provides a free tier via the Bailei model service, offering up to 1 million free tokens for the DeepSeek‑R1 full‑feature model. Deployment typically takes about five minutes with no cost during the trial period.

Steps:

Open the Bailei model service portal and click “Free Experience”.

Agree to the agreement and claim the free quota.

Select the desired DeepSeek model version.

Confirm activation; the service automatically handles load balancing, auto‑scaling, and provides a visual Chatbox client.

After completing these steps the model is ready for use; testing shows low latency and stable operation.