When to use Kumo Agent
Use Kumo Agent when you know the task but do not know which model should handle it. It can route simple chat to lighter models and delivery tasks to stronger routes.
Kumo Lumo
Pick the route that matches the task.
Kumo Lumo makes model choice easier by grouping routes around real work: chat, work delivery, difficult reasoning, image generation, voice workflows and video planning.
| User goal | Suggested route | Why | Point profile | Access note |
|---|---|---|---|---|
| Daily chat | GPT-5.4-Mini or Gemini-3.5-Flash | Fast enough for simple questions and low-cost replies | Low | Good for lightweight usage |
| Work delivery | Kumo Agent or GPT-5.4 | Balances output quality, structure and cost | Medium | Member route for heavier work |
| Difficult task | Claude-Opus-4.8 or Gemini-3.1-Pro | Better for long context, careful reasoning and synthesis | High | Premium route with daily previews for lower plans |
| Image generation | GPT Image-2 or Nano-Banana-Pro | Builds visual specs, prompts and image generation directions | Variable | Media workflow access required |
| Video generation | Seedance, Kling, Veo, Sora or Runway | Good for scene lists, motion, duration and video prompt planning | Medium to high | Higher tiers unlock stronger routes |
When to choose manually
Choose manually when you already know the model family you want, such as a long-context model for documents or an image route for visual output.