AI Models

Open in CMS

The AI Models page decides which models your organization's agents can use, where those models come from, and which model powers each platform feature by default.

Open the Organization page → Configure on your organization → AI Models tile. The tile is shown only to Owners and Admins.


Three tabs

TabWhat you do there
ModelsEvery model the organization can use, grouped by source, each with an on/off switch. A model switched off is hidden from every agent.
SourcesConnect your own sources: Gateways and API keys. Shown only where Qlar has allowed at least one of them for your organization.
DefaultsChoose the model that powers each feature — Conversation, Summarization, Personalization, Reranker, KB Training and Conversation Analysis. Leave one unset to use the Qlar default.

Where models come from

SourceWhat it isQlar creditsSet it up
QlarBuilt-in models, ready to use.YesNothing to set up.
API keyA model on your own OpenAI, Azure OpenAI, Azure AI Inference or DeepSeek account, reached with your key.None for the modelBring Your Own Key
GatewayYour own Claude Code or Codex (ChatGPT) account — subscription or API key — running on a computer you control.None for the modelConnect a Gateway

API keys and gateways are turned on per organization by Qlar. If you'd like to use them, ask Qlar support. Tools an answer uses — web search, knowledge base, plugins — are billed in credits whatever the model's source.

Which to choose? Start with Qlar models. Add an API key when you already have a provider contract or need a specific model. Connect a gateway when you want agents to use a Claude or ChatGPT plan you already pay for, and a computer of yours can stay on; a gateway runs 2 requests at a time, so it suits internal teams and pilots.


Who can do what

RoleAI Models
Owner / AdminOpen the page, switch models on or off, add keys, connect gateways, and set defaults.
BuilderCan't open the page, but picks from the organization's models when configuring an agent.
OthersNo access.