Parameters and Extra Features

Chat Configuration

For fine-grained control over your conversations:

• Click the gear icon next to the model picker in the input bar
• Adjust parameters like temperature and top_p
• Set a custom system prompt or select one from your prompt library (configurable in Settings)
• Press save to save changes

Provider Options

The chat config settings include a section for Provider Options (JSON) where you can pass additional parameters to model providers:

• These options are passed directly in the body of API requests
• Must be valid JSON. No trailing commas but quotes can be omitted for keys
• Press save to save changes
• Refer to Vercel AI SDK Providers documentation and your provider's API docs for available options

Provider Options Examples

For Gemini image generation:

{
  "response_modalities": ["TEXT", "IMAGE"]
}

For Claude extended thinking:

{
  "thinking": {
      "type": "enabled",
      "budget_tokens": 12000
    }
}