Getting Started with DeepChat: Install, Configure, and Connect AI Models

Setting Up DeepChat in 10 Minutes — A Real Walkthrough

I'm going to show you exactly how I got DeepChat running with both cloud models (GPT-4) and local models (via Ollama). No fluff, just the steps that actually work.

Step 1: The Quick Docker Install

This is the fastest way to get started. Run this in your terminal:

docker run -d \
  --name deepchat \
  -p 3000:3000 \
  -v deepchat-data:/data \
  luzixu/deepchat:latest

Takes about 30 seconds to pull (the image is ~450MB). Once it's running, open http://localhost:3000 in your browser.

What you should see: A clean chat interface with a setup wizard. If you see a blank page, wait 10 more seconds — the first start can be slow as it initializes the database.

Step 2: Connecting AI Models

Here's where I wasted 30 minutes. The settings panel isn't obvious at first — look for the gear icon in the top-right corner.

Adding Cloud Models (GPT-4)

1. Go to Settings → Model Providers
2. Click "Add Provider" → OpenAI
3. Paste your API key (you'll need one from platform.openai.com)
4. Select GPT-4 or GPT-3.5-turbo from the dropdown
5. Click Save

Adding Local Models (Ollama)

1. Make sure Ollama is running locally on your machine
2. In DeepChat settings, add a custom provider
3. Set the endpoint to: http://host.docker.internal:11434 (or your Ollama host)
4. Choose any model you've pulled (e.g., llama3.2, mistral)
5. Save and switch to it — the response times are impressive for local hardware

Step 3: Connect Your Knowledge Base

This is what sold me on DeepChat. Go to Knowledge → Add Source. You can point it to a local folder, upload files, or connect to Notion. I pointed it at my ~/Documents/notes/ folder. It indexed ~50 markdown files in under a minute.

Pro tip: Use the RAG mode when asking questions about your documents. The model will search your indexed files and reference specific sections. I tested it with a 200-page research paper — it found the relevant section in under 2 seconds.

Common Gotchas

• Blank Docker screen? The container needs a few seconds to initialize. Check docker logs deepchat — if you see "Listening on port 3000," you're good.
• Ollama not connecting? If you're on macOS, Ollama runs on localhost:11434 by default. From Docker, use host.docker.internal instead of localhost.
• API key errors? Make sure there are no trailing spaces when pasting keys into DeepChat's settings.

Final Verdict

DeepChat isn't the most polished AI assistant out there, but it's the most capable open-source desktop option for multi-model + private knowledge base use. If you value privacy and want to use multiple AI models without juggling tabs, this is worth your time.