AI Model Selector - VS Code Extension
A VS Code Chat provider for custom OpenAI-compatible models. Define your endpoints and models in a simple JSON config file, then use the extension to surface them in VS Code Chat.
Features
- Custom model provider for VS Code Chat
- Auto-loads models from
.vscode/model.json
- Auto-refreshes when the config file is saved
- Supports OpenAI-compatible endpoints and self-hosted APIs
- Workspace config with global fallback for no-workspace scenarios
Quick Start
- Open the command palette (
Ctrl+Shift+P / Cmd+Shift+P)
- Run
ai-model: Config
- Edit the generated
.vscode/model.json
- Save the file and open VS Code Chat
Config File
The extension stores model definitions in .vscode/model.json for workspace use. If no workspace is open, it falls back to a model.json file in the extension's global storage.
The config file is created automatically on first activation if it does not exist.
Example config
{
"models": [
{
"id": "deepseek-v4-pro",
"apiKey": "sk-your-api-key",
"baseUrl": "https://api.deepseek.com/v1",
"maxInputTokens": 128000,
"maxOutputTokens": 16000
}
]
}
Field Reference
| Field |
Description |
id |
Model identifier used internally and in the picker |
name |
Display name shown in VS Code Chat |
vendor |
Optional vendor identifier, inferred from id if omitted |
apiKey |
API key for the endpoint |
baseUrl |
Base URL for API requests (example: https://api.deepseek.com/v1) |
maxInputTokens |
Maximum input token count |
maxOutputTokens |
Maximum output token count |
imageInput |
Optional boolean to indicate image input support |
toolCalling |
Optional boolean to indicate tool calling support |
Compatibility
The extension also recognizes legacy endpoints config format and converts it to the current models format automatically.
Commands
| Command |
Description |
modelSelector.config |
Open the model configuration file |
Development
npm install
npm run compile
# then press F5 in VS Code to launch the extension host
Packaging
npm install -g @vscode/vsce
vsce package
License
MIT License
