Intelligence

Bring your own key, your own model

Nolocron doesn't marry one model vendor. Bring your own keys and point the assistant at the provider and model you prefer; swap any time. The model only ever sees the slice you granted.

AvailableModels

Same question, two windows

Small local model8K window
Large cloud model200K window
Output reserveSystemYour messageTool catalogHistoryRetrievalDerived
On the small model, History and Retrieval get squeezed first — and whatever is trimmed is noted on the run’s receipt.
01

Tie your memory to one model and you inherit its pricing, its limits, and its lifespan. The substrate should outlive any single model.

02
  • Bring-your-own-key across OpenAI, Anthropic, Google, and xAI.
  • Keys live in your system's secure storage and reload without a restart.
  • Capability-aware model selection for chat, structured output, and tool use.
  • Per-run cost tracking, so you can see what a model is costing you.
03
  1. Add a provider key in settings.
  2. Pick the model the assistant should use.
  3. Run your work; switch providers whenever you like.
  4. Watch per-run cost as you go.
04
  • No lock-in at the model layer.
  • Keys stay in secure local storage.
  • Models come and go; your archive is the part that stays.