AI Services

Calvin CS provides a local AI system for students and faculty to access and test local open-source LLM AI models.

http://localai.calvin.edu

This is an instance of OpenWebUI, which provides a web interface for using and managing local AI models. CS runs a few GPU-capable servers that host models for this system. These models will rotate seasonally, usually once in the summer, once in the fall, and once in the spring. The current models available are:

redhatai-qwen3-6-35b-a3b-nvfp4
google-gemma-4-26B-A4B-it

VS Code Setup instructions

Pre-requisites:

VS Code Insiders (https://code.visualstudio.com/insiders/) ** NOTE: As of 2026-06-05, only VS Code Insiders is supported for custom AI agents. **
VS Code Extension: "GitHub Copilot Chat", which you have logged in with your Github account.
Previously logged-in OpenWebUI LocalAI account (https://localai.calvin.edu/)

Steps:

Log into OpenWebUI LocalAI (https://localai.calvin.edu/) and create an API key for VS Code access.
- Click on the user icon in the top right corner and select "Settings". View screenshot
- Click "Settings". View screenshot
- Click "Account". View screenshot
- Show the API keys section. View screenshot
- Generate your API key and copy it to your clipboard.
- Make sure to hit "Save" after generating the API key, or it will be lost when you navigate away from the page.
Open VS Code Insiders.
- Open the Command Palette (Ctrl+Shift+P or Cmd+Shift+P).
- Search for "Chat: Manage Language Models" and select it.
- Click "Add Models" -> "Add Custom Endpoint". View screenshot
- Name: your choice, but a suggestion is "CS LocalAI".
- API Key: paste the API key you copied from OpenWebUI LocalAI.
- API Type: Chat Completions API.
- You will need to fill in the information for each model manually. The models that are currently available can be found on the top of the page, or by visiting https://localai.calvin.edu/api/v1/models. An example model is this:
```
    "models": [
        {
            "_modelInformationUrl": "https://localai.calvin.edu/api/v1/models",
            "id": "redhatai-qwen3-6-35b-a3b-nvfp4",
            "name": "Qwen3.6-35B-A3B",
            "url": "https://localai.calvin.edu/api/v1/",
            "toolCalling": true,
            "vision": true,
            "maxInputTokens": 128000,
            "maxOutputTokens": 16000,
            "completionOptions": {
                "stream": false
            }
        }
    ]
```
  For each model, you will need to fill in the following information:
  - Model Name: This can be whatever you want, but a suggestion is to use the "name" field from the API, such as "Qwen3.6-35B-A3B".
  - Model ID: This needs to match the "id" field from the API, such as "redhatai-qwen3-6-35b-a3b-nvfp4".
  - Endpoint URL: This should be "https://localai.calvin.edu/api/v1/".
- If you ever need to add more models or edit existing ones, use the Command Palette (Ctrl+Shift+P or Cmd+Shift+P) and select "Chat: Open Language Models (JSON)".