Understand when to use Reasoning, Creative, Code, or Vision models for the best results.
Vigthoria separates dispatch, architecture, execution, and realtime assistance so every request uses the right capability without loading unnecessary context.
Multimodal architecture, planning, research, image understanding, and long-video analysis.
Fast chat, completion, diagnostics, regex, and focused single-step work.
Complete implementation, repository refactoring, deep debugging, and security work.
Native image, screenshot, diagram, and long-video understanding on the Creative architect lane.
| Capability | Reasoning | Creative | Code | Vision |
|---|---|---|---|---|
| Complex Analysis | ★★★★★ | ★★★☆☆ | ★★★★☆ | ★★★☆☆ |
| Creative Writing | ★★★☆☆ | ★★★★★ | ★★☆☆☆ | ★★★☆☆ |
| Code Generation | ★★★★☆ | ★★☆☆☆ | ★★★★★ | ★★★☆☆ |
| Image Understanding | — | — | — | ★★★★★ |
| Speed | Medium | Fast | Fast | Medium |
| Context Window | 128K | 128K | 128K | 64K + Images |
You need engaging, well-structured content that captures readers' attention.
You have a tricky bug that requires understanding multiple systems and trace logic.
You need to write new code for a specific feature with best practices.
You want AI to review a UI mockup, diagram, or error screenshot.
You need to analyze market data and create a strategic plan.
Use Creative v4.2 for planning and visual evidence, then hand the approved blueprint to Code v4.2 for implementation.
Set your most-used model as default in Settings → AI → Default Model. You can always override per conversation.
If you are unsure, leave routing on automatic: Olivia selects the toolkit and the platform applies the correct architect, executor, or realtime role.
