Ollama Setup on Windows: Install and Run Qwen 3.8 Locally (No GPU Needed), by Laxmi Kant | KGP Talkie on YouTube. 8 min 24 s.
- Video length
- 8 min 24 s
- Time to do it
- 15 minutes
- Difficulty
- Easy
- Steps
- 9
What you need
Steps
Open your browser and go to ollama.com. Click the Download button and choose Windows to get the installer file.
Run the downloaded installer and accept the default options. No admin rights or account are needed, so the setup finishes in under a minute.
Check your RAM before picking a model. Qwen 3.5 4B needs about 3.1 GB, Gemma 4 12B about 7.5 GB, and Qwen 3.8 27B about 16.5 GB.
Open the Ollama app and start a chat with Qwen 3.5 4B. Type a short prompt and press Enter to see the reply appear on your screen.
Open PowerShell and run ollama run qwen3.5:4b to chat from the terminal. Run ollama ps to see which model is loaded and how it is running.
If you have no graphics card, stick with the 4B model. It runs on the processor alone at roughly 18 tokens per second, which is fine for chat.
Get a second model with ollama pull gemma4:12b. Wait for the download to finish, then run it the same way you ran the first model.
Open the Settings page to turn auto updates on or off, allow or block network access, and move the model folder to another drive. Set the context length there too.
Remove models you no longer use with ollama rm followed by the model name. Turn off auto-start if you do not want Ollama running at login.
Tips and common mistakes
Pick a model that fits your RAM with room to spare, or the chat will slow down badly.
Keep the model folder on a drive with plenty of free space, since each model can take several GB.
Use ollama ps to confirm whether a model is running on the GPU or the CPU before blaming slow replies.
Block network access in Settings if you want to be sure nothing leaves your machine.
Questions
Do I need a graphics card to run Ollama on Windows?
No. A small model like Qwen 3.5 4B runs on the processor alone at about 18 tokens per second. Larger models will be much slower without a GPU.
Where are Ollama models stored?
By default they sit in your user folder on the C drive. You can change the model location to another drive on the Settings page.
Can other apps use my local Ollama model?
Yes. Ollama serves a local address that other apps can connect to. Check the Settings page for that address and for the network access option.
This guide was drafted from the video's details and general task knowledge. It has not been checked against the full video. Check the video and relevant product instructions before following a procedure. Report a correction.
