- Overview
- Installation and setup
- Getting started
- Trust and compliance
- Governance
- Features
- Reference
- Troubleshooting
- Backup and restore
Choosing the LLM model
Select which AI model processes your requests in Cartographer, from the models your organization has enabled.
Select the model from the dropdown next to the chatbox — the AI model that processes your requests, set by your organization by default:
Selected model
The model determines Cartographer's reasoning quality, speed, and capabilities. Which models are available, and which one is used by default for new conversations, is set by your organization — see Governance with Automation Ops — and depends on your UiPath tenant configuration. You can change the model for your own session if the model you want is enabled.
Stronger models for agents
Below the model list in the picker, a Stronger models for agents toggle (off by default) lets Cartographer's agent actions use a more capable model than the one selected above for your chat — "May cost more" shows next to the toggle as a reminder.
Available models
For the Cartographer skills, an LLM in the category of GPT 5.6 Luna with very high effort or better is a minimum. GPT 5.6 Terra (or similar) is recommended.
On the Enterprise plan, Cartographer is free on GPT-5.6 Luna as a trial offer until November 23, 2026. Usage on all other billable models counts toward your monthly quota. On the Community and Enterprise Trial plans, Cartographer always draws from the usage pool.
| Provider | Models |
|---|---|
| Anthropic (Claude) | Claude Sonnet 5, Claude Opus 4.8 |
| Azure OpenAI | GPT-5.6 Sol, GPT-5.6 Terra, GPT-5.6 Luna |
| Google (Gemini) | Gemini 3.6 Flash |
| Moonshot AI | Kimi K3 |
Your organization can also bring its own model instead of using the models listed above — see Configuring LLMs for how to set this up.
A model must be enabled both in AI Trust Layer → Models and through an Automation Ops policy to be usable in Cartographer.
AI Trust Layer is the source of truth.
A model can be allowed by policy. If its provider is disabled in AI Trust Layer, it is still not available.
See also AI Trust Layer → PII & Guardrails and AI Trust Layer → Regions for related tenant-level controls.