Uploaded October 2026 | Updated November 2026, 2 days ago
Install Ollama on Windows and run your own private AI, free, with nothing leaving your computer. Day 1 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
Next, day 2, DeepSeek Harness: youtu.be/QqxOEu7Cn7Q
What you need: Windows 10 or 11, about 4 GB free plus room for models. No account, no admin rights, and no graphics card required.
In this video:
- Download and install Ollama from ollama.com
- Pick a model by memory: Qwen 3.5 4B (3.1 GB), Gemma 4 12B (7.5 GB), Qwen 3.8 27B (16.5 GB)
- First chat in the Ollama app
- Chat in PowerShell with ollama run, and check the hardware with ollama ps
- The no graphics card test: Qwen 3.5 4B on the processor only, about 18 tokens per second
- Get a second model from the library with ollama pull gemma4:12b
- Gemma 4 reads a screenshot of a real PowerShell error, fully offline
- The Settings page: auto updates, network access, model location on another drive, context length
- Remove models with ollama rm, stop auto-start, and the local address other apps use
Chapters
0:00 My own AI writing an email, offline
0:21 Day 1 of 30: what Ollama is
0:45 What you need
1:12 Step 1: Download Ollama
1:31 Step 2: Install
1:55 Step 3: Pick a model for your memory
2:40 Step 4: First chat with Qwen 3.8
3:20 Step 5: Ollama in PowerShell
3:57 Step 6: No graphics card (CPU only)
4:30 Step 7: A second model with ollama pull
5:29 Step 8: Gemma 4 reads a screenshot
6:04 Step 9: Ollama settings and context length
7:09 Three tips that save time
7:49 Ollama vs ChatGPT Plus
8:09 Tomorrow: day 2
Links
Ollama download: ollama.com/download
Qwen 3.8 speed test on an RTX 5090: kgptalkie.com/tutorials/llm-benchmarking/qwen-3-8-27b-llama-cpp-speed-settings
Tomorrow, day 2: the DeepSeek Harness desktop app.
New free AI tool every day. Subscribe so you don't miss one: youtube.com/@KGPTalkie?sub_confirmation=1
Install Ollama on Windows and run your own private AI, free, with nothing leaving your computer. Day 1 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
Next, day 2, DeepSeek Harness: youtu.be/QqxOEu7Cn7Q
What you need: Windows 10 or 11, about 4 GB free plus room for models. No account, no admin rights, and no graphics card required.
In this video:
- Download and install Ollama from ollama.com
- Pick a model by memory: Qwen 3.5 4B (3.1 GB), Gemma 4 12B (7.5 GB), Qwen 3.8 27B (16.5 GB)
- First chat in the Ollama app
- Chat in PowerShell with ollama run, and check the hardware with ollama ps
- The no graphics card test: Qwen 3.5 4B on the processor only, about 18 tokens per second
- Get a second model from the library with ollama pull gemma4:12b
- Gemma 4 reads a screenshot of a real PowerShell error, fully offline
- The Settings page: auto updates, network access, model location on another drive, context length
- Remove models with ollama rm, stop auto-start, and the local address other apps use
Chapters
0:00 My own AI writing an email, offline
0:21 Day 1 of 30: what Ollama is
0:45 What you need
1:12 Step 1: Download Ollama
1:31 Step 2: Install
1:55 Step 3: Pick a model for your memory
2:40 Step 4: First chat with Qwen 3.8
3:20 Step 5: Ollama in PowerShell
3:57 Step 6: No graphics card (CPU only)
4:30 Step 7: A second model with ollama pull
5:29 Step 8: Gemma 4 reads a screenshot
6:04 Step 9: Ollama settings and context length
7:09 Three tips that save time
7:49 Ollama vs ChatGPT Plus
8:09 Tomorrow: day 2
Links
Ollama download: ollama.com/download
Qwen 3.8 speed test on an RTX 5090: kgptalkie.com/tutorials/llm-benchmarking/qwen-3-8-27b-llama-cpp-speed-settings
Tomorrow, day 2: the DeepSeek Harness desktop app.
New free AI tool every day. Subscribe so you don't miss one: youtube.com/@KGPTalkie?sub_confirmation=1










