Meridian Line's legal team has a long-standing enterprise agreement with a third-party model provider outside Databricks, and wants Meridian's charter-desk assistant to call that provider's model while still getting Databricks-side rate limiting, usage tracking, and a single governed endpoint. What should the engineer configure?
Select an answer to reveal the explanation.
Short Explanation
External Models is the adapter for this setup: you route your assistant through one Databricks endpoint, which proxies requests to the outside provider while applying Databricks-side rate limits and usage tracking. That lets you use the provider you already have an agreement with and keep a governed access point.
Full Explanation
An External Models endpoint in Model Serving is designed for this setup: it proxies requests to a supported model hosted by a third-party provider outside Databricks, while keeping the endpoint in Databricks and allowing AI Gateway controls such as rate limiting and usage tracking. Your assistant can therefore call one governed endpoint without the model needing to run on Databricks. A Foundation Model APIs pay-per-token endpoint serves Databricks-hosted foundation models; entering an outside provider’s model name does not make Databricks proxy that provider’s model. A Direct Vector Access index is for retrieval from vector data, not for invoking a chat or completion model, and configuring it with the provider’s API key does not turn it into a model endpoint. A Delta Sync index is also a retrieval component, not a way to call a provider’s hosted model. Exam caveat: External Models applies to supported providers and is distinct from serving Databricks-hosted foundation models. As a practical check, verify that the provider and model are supported, configure the provider credentials and desired AI Gateway controls, then route the assistant to the resulting Model Serving endpoint.