Skip to main content
When you send mode: "auto" (the default), Theo classifies your prompt to determine the optimal execution mode. You can also set the mode explicitly to skip classification entirely.

​
Available Modes

|| Mode | Description | Engine | ||------|-------------|--------| || auto | Theo classifies automatically | Varies by classification | || fast | Quick responses, simple tasks | theo-1-flash | || think | Deep reasoning, complex analysis | theo-1-reason | || code | Code generation and engineering | theo-1-code | || image | Image generation | theo-1-create | || video | Video generation (async) | theo-1-motion | || research | Web research with citations (async) | theo-1-research | || roast | Humorous, unfiltered responses | theo-1-edge | || genui | Generative UI components | theo-1-genui |

​
Classification Process

​
How Classification Works

Theo uses a multi-tier classification system. Obvious signals are caught instantly:
  • "Draw me a logo" → image
  • "Write a React component" → code
  • "Research the latest AI news" → research
Ambiguous prompts are analyzed more deeply. The classifier returns: 
{ "mode": "think", "confidence": 0.92, "reasoning": "Complex analytical question requiring deep reasoning" }

​
Explicit Override

Pass mode directly to skip classification: 
const res = await theo.complete({
  prompt: "Write a REST API for user management",
  mode: "code",  // Skips classification entirely
});

​
Mode Selection in the Response

Every response includes both the requested mode and the resolved mode: 
{
  "mode": "auto",
  "resolved_mode": "code",
  "model": { "id": "theo-1-reason", "label": "Theo Reason" }
}