Models & Providers
Manage which AI models power your assistants and how providers are configured. Admin access required.
The Page
Go to Models & Providers. Three tabs:
| Tab | What It's For |
|---|---|
| Models | The model catalog — add, edit, set defaults |
| Provider Keys | API keys for external providers |
| Tenant Configs | Per-organization provider status |
Models
Each registered model shows its status, a Default badge if it's the default, and its capacity limits.
Adding a Model
- Click Add Model
- Fill in:
- Model Name and Display Name
- Provider — the platform's own model, or an external provider (OpenAI, Anthropic, Google Gemini)
- Endpoint URL — filled automatically for external providers
- Save
caution
External providers show a warning: conversation data leaves your infrastructure when using them. The platform's own model keeps everything in-house.
Editing a model also lets you toggle Active and Default.
Provider Keys
Manage API keys for external providers:
- Add API Key — give it a Label, paste the key, and optionally assign it to a specific organization (or leave as default for all)
- Test — verify a key works
- Delete — remove a key
Keys are masked after saving.
Tenant Configs
A status matrix of organizations × providers, with states like Active, Error, Pending, or Disabled. Open an organization to see, per provider:
- The enable toggle and current status
- Which key is in use
- Doc Sharing — whether the AI may use your documents with this provider
- Test Connection — verify it's working
Tips
- Test after changes — always confirm a provider works before relying on it
- Keep a default model active — assistants need at least one working model
- Prefer the platform model for sensitive data — external providers receive conversation content
- Rotate keys periodically