Setting Up Ollama (Power User)

Snipset includes powerful AI capabilities, such as AI Chat, AI Snippet Generation, and Semantic Search (OmniSearch). This guide covers the Advanced, CPU-optimized installation for power users who want to save disk space and don't need GPU acceleration.

Why use the CPU-Optimized version?

Ollama detects GPU availability dynamically. If the GPU libraries are removed, Ollama automatically falls back to the CPU using the ggml-cpu-*.dll library. Since Snipset uses cloud models and lightweight embedding models, you don't need local GPU processing. Removing these libraries saves up to ~1.75 GB of storage space.

Step 1 — Install Custom Ollama

1. Download the ZIP Version Go to Ollama GitHub Releases , find the latest version, and download the ollama-windows-amd64.zip file.
2. Extract the Folder Extract the zip file and save the folder in your Local Disk C (e.g., C:\ollama-windows-amd64).
3. Remove GPU Libraries (Storage Optimization) Open PowerShell and run the following commands to remove the GPU folders (saving ~1.75 GB):

    Remove-Item -Recurse -Force "C:\ollama-windows-amd64\lib\ollama\cuda_v12" Remove-Item -Recurse -Force "C:\ollama-windows-amd64\lib\ollama\cuda_v13" Remove-Item -Recurse -Force "C:\ollama-windows-amd64\lib\ollama\vulkan" 

   Alternative for WSL users:

    rm -rf "/mnt/c/ollama-windows-amd64/lib/ollama/cuda_v12" rm -rf "/mnt/c/ollama-windows-amd64/lib/ollama/cuda_v13" rm -rf "/mnt/c/ollama-windows-amd64/lib/ollama/vulkan" 

   (The remaining size of the Ollama folder is now only about 150 MB.)
4. Register to Environment Variable (Path) So your computer can detect the ollama command: _ Open the Start Menu, search for Environment Variables and select "Edit the system environment variables". _ Click the Environment Variables... button. _ Under System variables or User variables, find the Path variable, then click Edit... _ Click New and enter your folder path: C:\ollama-windows-amd64 * Click OK on all windows to save.

Step 2 — Pull the Required Models

Open your terminal and pull the desired models:

Model 1: The Chat Model

Model 2: The Embedding Model

Step 3 — Verify Ollama is Running

• Open your web browser and navigate to http://localhost:11434
• If Ollama is running correctly, you will see: "Ollama is running".

Step 4 — Connect Snipset to Ollama

1. Go to Snipset → Settings → AI Configuration tab.
2. Ensure Base URL is http://localhost:11434.
3. Verify the green Online indicator.
4. Select your Chat Model and Embed Model from the dropdowns.