Uploaded October 2026 | Updated November 2026, 4 days ago
Install Open WebUI on Windows and get your own ChatGPT-style app in the browser, running on your own computer with Ollama. Chat with local models and ask questions about your own PDFs, with nothing sent to the cloud. Day 5 of the Free AI Setup series: one free AI tool a day, set up step by step.
Watch the whole series in order: youtube.com/playlist?list=PLfUBzZXRNC-k
Previous, day 4, LM Studio: youtu.be/396Xg5ZSL80
Next, day 6, OpenCode: youtu.be/SnkjLuvkh2U
What you need: Windows 10 or 11, Ollama with at least one model (day 1 of this series) and about 2 GB of free disk space. No Docker, no separate Python install and no online account. A graphics card is optional: without one, pick a small model like Qwen 3.5 4B.
In this video:
- The two install routes: Docker (the docs' pick for most users) and uv (the easy route on Windows)
- Install uv with one PowerShell line
- Why uv: Open WebUI needs Python 3.11 or 3.12, and uv brings the right one
- Start Open WebUI with uvx and a data folder that keeps your chats safe
- First start: 245 packages, about 1.7 GB, about 1 minute 40 seconds
- Fix "Cannot GET /" on port 8080: another app was using it, so start on port 3000
- Create the local admin account (stored only on your computer)
- Ollama connects by itself: where to check it in Admin Panel, Settings, Connections
- Stop the red "Missing bearer authentication" errors by switching off the unused OpenAI API connection
- Pick a model, set it as default, and download new Ollama models from the model picker
- First chat with Qwen 3.8 on an RTX 5090, about 114 tokens per second
- Upload a PDF and ask about it: answers with sources you can check, in under 3 seconds
- Switch to Gemma 4 12B for the same document
- Tips: start it again, stop it, open it from your phone, stay on the latest version
Chapters:
0:00 Intro: your own ChatGPT on your computer
0:28 Day 5 of 30 and what Open WebUI is
0:57 What you need
1:30 Two ways to install: Docker or uv
2:07 Step 1: Install uv
2:38 Why uv brings Python 3.11
3:01 Step 2: Start Open WebUI
4:05 Step 3: Fix Cannot GET on port 8080
4:56 Step 4: Create the admin account
5:47 Step 5: Ollama connects by itself
6:22 Switch off the OpenAI connection
6:52 Step 6: Pick a model and set the default
7:34 Step 7: First chat with Qwen 3.8
8:33 Step 8: Chat with your own PDF
9:58 Step 9: Switch to Gemma 4 12B
10:34 Four quick tips
11:25 Open WebUI vs ChatGPT Plus
11:59 Next: OpenCode
Links
Open WebUI docs (Quick Start): docs.openwebui.com/getting-started/quick-start
Open WebUI on GitHub: github.com/open-webui/open-webui
uv: https://docs.astral.sh/uv/
Ollama (day 1 of this series): ollama.com/download
Tomorrow, day 6: OpenCode.
New free AI tool every day. Subscribe so you don't miss one: youtube.com/@KGPTalkie?sub_confirmation=1
Install Open WebUI on Windows and get your own ChatGPT-style app in the browser, running on your own computer with Ollama. Chat with local models and ask questions about your own PDFs, with nothing sent to the cloud. Day 5 of the Free AI Setup series: one free AI tool a day, set up step by step.
Watch the whole series in order: youtube.com/playlist?list=PLfUBzZXRNC-k
Previous, day 4, LM Studio: youtu.be/396Xg5ZSL80
Next, day 6, OpenCode: youtu.be/SnkjLuvkh2U
What you need: Windows 10 or 11, Ollama with at least one model (day 1 of this series) and about 2 GB of free disk space. No Docker, no separate Python install and no online account. A graphics card is optional: without one, pick a small model like Qwen 3.5 4B.
In this video:
- The two install routes: Docker (the docs' pick for most users) and uv (the easy route on Windows)
- Install uv with one PowerShell line
- Why uv: Open WebUI needs Python 3.11 or 3.12, and uv brings the right one
- Start Open WebUI with uvx and a data folder that keeps your chats safe
- First start: 245 packages, about 1.7 GB, about 1 minute 40 seconds
- Fix "Cannot GET /" on port 8080: another app was using it, so start on port 3000
- Create the local admin account (stored only on your computer)
- Ollama connects by itself: where to check it in Admin Panel, Settings, Connections
- Stop the red "Missing bearer authentication" errors by switching off the unused OpenAI API connection
- Pick a model, set it as default, and download new Ollama models from the model picker
- First chat with Qwen 3.8 on an RTX 5090, about 114 tokens per second
- Upload a PDF and ask about it: answers with sources you can check, in under 3 seconds
- Switch to Gemma 4 12B for the same document
- Tips: start it again, stop it, open it from your phone, stay on the latest version
Chapters:
0:00 Intro: your own ChatGPT on your computer
0:28 Day 5 of 30 and what Open WebUI is
0:57 What you need
1:30 Two ways to install: Docker or uv
2:07 Step 1: Install uv
2:38 Why uv brings Python 3.11
3:01 Step 2: Start Open WebUI
4:05 Step 3: Fix Cannot GET on port 8080
4:56 Step 4: Create the admin account
5:47 Step 5: Ollama connects by itself
6:22 Switch off the OpenAI connection
6:52 Step 6: Pick a model and set the default
7:34 Step 7: First chat with Qwen 3.8
8:33 Step 8: Chat with your own PDF
9:58 Step 9: Switch to Gemma 4 12B
10:34 Four quick tips
11:25 Open WebUI vs ChatGPT Plus
11:59 Next: OpenCode
Links
Open WebUI docs (Quick Start): docs.openwebui.com/getting-started/quick-start
Open WebUI on GitHub: github.com/open-webui/open-webui
uv: https://docs.astral.sh/uv/
Ollama (day 1 of this series): ollama.com/download
Tomorrow, day 6: OpenCode.
New free AI tool every day. Subscribe so you don't miss one: youtube.com/@KGPTalkie?sub_confirmation=1










