K
Kibo

AI Models

Kibo supports two local AI models. Choose the one that fits your needs.

Model Comparison

Needle 2

Size14 MB
SpeedFast
RAM UsageLow
Best ForQuick commands
AuthorCactus Compute

FunctionGemma

Size253 MB
SpeedModerate
RAM UsageHigher
Best ForComplex reasoning
AuthorGoogle DeepMind

Switching Models

You can switch models at any time during a session:

needle # Switch to Needle 2 gemma # Switch to FunctionGemma models # Show current model

When to Use Each

Use Needle 2 when:
  • You need fast responses
  • You're on a low-RAM machine
  • You're running simple commands (ls, battery, volume)
  • You want the lightest footprint
Use FunctionGemma when:
  • You need complex reasoning
  • You're working with ambiguous commands
  • You need better tool selection
  • You have enough RAM available

Configuration

Your model preference is stored in:

~/.agent_model_config.json

This file is automatically updated when you switch models.

CommandsTools