Skip to main content
The LLM catalog is the Model Catalog: a single pane over every AI provider and model in your organisation. It replaces raw API keys and scattered environment variables with governed Provider Integrations, so you can see what is available, who can use it, and what it costs.

Model Catalog

Providers, models, and the @provider/model slug you use in code

Gateway Models

The models the AI Gateway serves and how they are named

​
Getting models into the catalog

An Integration connects the AI Gateway to a provider — OpenAI, Anthropic, Bedrock, or a self-hosted endpoint. Creating one is what puts that provider’s models in the catalog, and it is also where you decide which workspaces see them and what they are allowed to spend.

LLM Integrations

Create an Integration and connect a provider

Custom Models

Add models the AI Gateway does not know about yet

​
Tuning what the catalog reports

Pricing and model metadata come from the provider by default. Where your contract differs, or where a model behaves differently in your deployment, override it.

Pricing Adjustments

Apply your negotiated rates so cost reporting is accurate

Model Overrides

Change token limits, capabilities, and other model details
Last modified on September 15, 2026