Models

Models connect your organization's agents to model providers. Let's add a provider, enable its models, and assign them to an agent.

Add a provider

  1. Open Models and choose Add provider.
  2. Give the connection a unique Name, such as OpenAI production. You can add multiple OpenAI connections with different keys or API URLs.
  3. Choose OpenAI, Anthropic, or OpenRouter, then set the API URL. OpenAI defaults to Responses and also offers Chat Completions for compatible gateways. Anthropic uses its Messages API. OpenRouter uses Chat Completions across its supported model providers. The provider is fixed once the connection is saved, so add another connection to use a different one.
  4. Enter the provider's API key. It is encrypted when saved. Later, the editor shows only the last four characters. Leave the key field blank to keep the stored key, or enter a replacement to rotate it. Changing the API URL requires entering the key again, because a stored key never follows its connection to another endpoint.
  5. Select the models to enable. All three providers offer catalog suggestions. For another model or private deployment, enter the exact Custom model ID and choose Add model. For OpenRouter, use the full model ID including its prefix, such as anthropic/claude-sonnet-4.6.
  6. Choose Save provider, then Set up an agent.

NOTE: Saving stores your configuration without making a model request. Catalog suggestions do not verify access, and your provider may restrict the models available to its key.

Assign models to an agent

  1. Open Agents and choose an existing agent or Add agent.
  2. Select one or more Models. Each choice includes its provider name, so the same upstream model can appear through multiple connections.
  3. Choose the Default model, then save the agent. New chat runs use that default through its provider's API URL and key.

The enabled provider models define what can be assigned to agents. Remove a model from its agents before disabling it on the provider or deleting the provider. Until you do, saving or deleting the provider fails with a message naming up to three of those agents. See Agents for the other agent settings.

Troubleshooting

  • If a model request fails, check the provider's API URL, API format, key access, and exact model ID. Tenant users see only that the assistant is unavailable, so the provider's own error never reaches them.
  • If saving refuses the API URL with Use a public HTTPS API URL. This server does not allow HTTP or private network endpoints, the deployment accepts only public HTTPS endpoints. A self-hosted deployment's operator can allow private endpoints in Configuration.
  • If the stored key cannot be read, enter it again in the provider editor and save.
  • If an agent has no model choices, enable a model on a provider first.
  • If a provider changed in another tab, reload its editor before saving again.