AI Providers

The plugin supports three provider modes, accessible from the Settings > AI Models tab.

Provider Modes

Free Mode

Two built-in free-tier slots with real AI models. No API key required, no setup needed. Just open the plugin and start prompting. See Free Tier for details.

AI Agent Mode

Run CLI-based AI agents directly inside the editor. Each agent runs as a background process with full access to all plugin tools via MCP. Supported agents:

• Claude Code (Anthropic)
• Codex (OpenAI)
• GitHub Copilot
• Gemini CLI (Google)

See AI Agents for setup instructions.

API Key / Local LLM Mode

Connect directly to cloud AI providers or local inference servers using your own API keys.

Supported Cloud Providers

Provider	Models	How to Get a Key
Gemini	Gemini Flash, Pro	Google AI Studio
OpenAI	GPT-4, GPT-4o, o1, o3	OpenAI Platform
Claude	Sonnet, Opus, Haiku	Anthropic Console
DeepSeek	DeepSeek Chat	DeepSeek Platform
Custom	Any model	Any OpenAI-compatible endpoint

Local LLM Support

Use any OpenAI-compatible local server. See Local LLM Setup.

API Key Slots

The plugin provides 9 independent API key slots. Each slot stores:

• Provider selection (Gemini, OpenAI, Claude, DeepSeek, Custom)
• API key
• Model name
• Custom base URL (for Custom provider)
• Custom request parameters (JSON)

Switching Slots

• Click the numbered slot buttons (1-9) in Settings > AI Models
• Hotkey switching: Press Left ALT + 1 through ALT + 9 to switch slots instantly from any view
• The active slot is shown with a colored indicator

Custom Provider Setup

For custom/local endpoints:

1. Select “Custom” from the Provider dropdown
2. Enter the Base URL (e.g., http://localhost:11434/v1/chat/completions)
3. Enter the Model Name (e.g., llama3.1)
4. API Key is optional for local servers
5. Custom Request Params (JSON) lets you add extra fields like {"temperature": 0.7, "max_tokens": 16384}

OpenRouter

OpenRouter gives you access to hundreds of models through a single API key:

1. Select “Custom” provider
2. Base URL: https://openrouter.ai/api/v1/chat/completions
3. Enter your OpenRouter API key
4. Model Name: any model from OpenRouter Models (e.g., anthropic/claude-sonnet-4)