Coming soon! The Kael'Nyrin Scrolls: The Atlas Edict

Captain Walker

Technology

Accessing DeepSeek V4-Pro via API on Windows 11 Using OpenWebUI

AI, API, deepseek, Docker, economical, local, OpenWebUI

Estimated reading time at 200 wpm: 6 minutes

The free DeepSeek chat platform at chat.deepseek.com is useful for general purposes, but it has limitations — rate limits, occasional capacity restrictions, and response quality that does not match what the paid API models can deliver. This guide explains how to access DeepSeek’s more capable models using OpenWebUI, a locally hosted chat interface that runs on your own machine. Conversations go directly from your machine to DeepSeek’s servers, with no third-party routing involved.

Whether or not you agree our Fat Disclaimer applies

Why Access DeepSeek Via the API?

DeepSeek has released its V4 model family — two models that represent a significant step forward in cost-effective AI. DeepSeek-V4-Pro carries 1.6 trillion total parameters (49 billion active), pre-trained on 33 trillion tokens, with performance described as rivalling the world’s top closed-source models. DeepSeek-V4-Flash, the lighter option, offers 284 billion total parameters (13 billion active) at a fraction of the cost of Gemini, Claude.ai, Grok etc.

Both models support a 1 million token context window — meaning they can hold an extraordinary amount of text in a single conversation. Both are open-source and available via API.

The API charges only for what you use, and the pricing is low. Output tokens on V4-Flash cost $0.28 per million; on V4-Pro, $0.87 per million. For context, a million tokens represents roughly 750,000 words — a substantial volume of professional work

By connecting to the DeepSeek API directly through a locally hosted interface, you get full model performance, far more headroom than the free platform allows, and a chat interface and history that live on your own machine — with costs that remain modest even under heavy use.

Known Limitation

Before proceeding, note one important limitation of this setup. No image uploads — DeepSeek’s API does not currently support image inputs. Attempting to upload an image will break the conversation thread. For image-related work, use a vision-capable model elsewhere.

Data Security and Confidentiality

Local hosting is not the same as privacy. The interface runs on your machine. Your chat history stays on your machine. But every prompt you send still travels to DeepSeek’s servers, and what happens to it there is governed by DeepSeek’s terms, not by you.

Two practical points follow.

First, your API key sits in plain text in the OpenWebUI admin panel. Anyone who can reach the interface can spend your credit. Keep port 3000 to the local machine. Do not expose it to your network or the internet without knowing exactly what you are doing.

Second, treat the whole arrangement as a third-party service for the purposes of confidentiality. Identifiable clinical, legal or personal material should not go through it.

I have written separately on this. See: [LINK — data security article]

What You Will Need

  • A Windows 11 machine
  • A DeepSeek account with API credit (platform.deepseek.com)
  • A DeepSeek API key: go to https://platform.deepseek.com/api_keys, create a key, and copy it. Deposit something small like $5 or $10 USD.
  • Docker Desktop for Windows
  • A browser

Step 1: Install WSL 2

Docker Desktop on Windows requires WSL 2 (Windows Subsystem for Linux). Open PowerShell as Administrator and run:

wsl --install

Follow the prompts. When complete, restart your machine before proceeding.

Step 2: Install Docker Desktop

Go to: https://docs.docker.com/desktop/install/windows-install

Download and run the installer. When the configuration screen appears, keep the defaults:

  • All-users installation selected
  • Use WSL 2 instead of Hyper-V ticked
  • Allow Windows Containers left unticked

Complete the installation and restart when prompted. When Docker Desktop opens, skip account creation — no Docker account is required.

Step 3: Install OpenWebUI

Run the following command in PowerShell:

docker run -d -p 3000:8080 --add-host=host.docker.internal:host-gateway -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:main

This downloads and starts OpenWebUI. It may take several minutes. If the download is interrupted, run the same command again — Docker will resume from where it left off.

If you encounter a conflict error on a second attempt, first run:

docker rm open-webui

Then run the original command again.

Step 4: Create Your Local Admin Account

Open your browser and go to: http://localhost:3000

You will be prompted to create an admin account. This is a local account only — it exists on your machine and makes no external connections. Enter any name, email address, and password. Click Create Admin Account.

Bookmark this address for future use.

Step 5: Connect DeepSeek API

In OpenWebUI, click your profile icon (bottom left) → Admin Panel → Settings → Connections.

Under OpenAI API, click the + button to add a new connection. Enter:

  • URL: https://api.deepseek.com/v1
  • API Key: paste your DeepSeek API key

Leave all other settings as default. Click Save.

Step 6: Start Chatting

Click New Chat. Select a model from the dropdown at the top:

  • deepseek-v4-flash — faster and cheaper; suitable for most tasks
  • deepseek-v4-pro — more capable reasoning; recommended for complex professional writing, research, and analysis

Watching the Cost

Two things quietly increase your spend.

The first is thread length. Every time you send a message, the entire conversation up to that point goes with it. A long thread therefore costs more per turn than a short one. Start a new chat when the subject changes.

The second is OpenWebUI itself. By default it makes extra API calls to generate a title and tags for each conversation. Small individually. Not small across hundreds of chats. To switch this off, go to Admin Panel → Settings → Interface and disable title generation and tag generation.

Check your usage at platform.deepseek.com from time to time.

Backing Up Your Chats

Chat history is stored locally on your machine inside the Docker volume. To back it up, go to your profile icon → Settings → Data Controls → Export Chats. Save the downloaded file to a cloud location such as OneDrive.

Do this periodically to avoid losing your conversation history.

Daily Use

From now on, to use OpenWebUI:

  1. Ensure Docker Desktop is running (check the system tray for the whale icon — it should start automatically with Windows)
  2. Open your browser and go to localhost:3000

Everything else is already configured and saved.