Nanobot works best when provider and model are explicitly set instead of relying on auto-detection.
Provider setup
Add Imole as an OpenAI-compatible provider, then point the primary model preset at that provider.
- 1
Set IMOLE_API_KEY in the environment.
- 2
Add an Imole provider with apiBase.
- 3
Select the provider in modelPresets.primary.
- 4
Choose a model from the catalog.
providers:
imole:
apiBase: https://api.imole.app/v1
apiKey: ${IMOLE_API_KEY}
modelPresets:
primary:
provider: imole
model: gpt-5.6-terraOpenAI API mode
Nanobot can also expose an OpenAI-compatible API for local integrations. That is the opposite direction: Nanobot becomes the server. Use the provider setup above when Nanobot should call Imole as the upstream model router.
Recommended models
Use gpt-5.6-luna for routine tests, gpt-5.6-terra for normal Agent work, and gpt-5.6-sol with high reasoning for coding. Other Agent reasoning defaults to medium.