Databricks

Access

Kind

A read-only key you paste

What to give

Workspace, token and a SQL warehouse

Why that one

A personal access token that can read the system tables and use one SQL warehouse. Valistry runs one query and nothing else.

What it attributes on its own

From the source

day; SKU as Databricks bills it; workspace; DBUs consumed; the list price of what was used, at the price that applied then; not your contract price

Needs a rule or a trace

the model and the tokens — a serving endpoint's bill is DBUs, not tokens, so neither is on these rows; team, workflow and agent unless a rule maps the SKU or the workspace; outcome

Connecting it

1. Ask an admin for billing access. Copy the grants. Send them to your metastore admin to run in the SQL editor. Open ↗
2. Create a token, copy warehouse ID. Your username → Settings → Developer → Access tokens → Manage → Generate new token

A serving endpoint's bill is in DBUs, so this source carries no tokens. If a gateway you connect also reaches these models, it reports the tokens and this reports the bill. Count the money from only one, or the same work counts twice.

Settings worth turning on in Databricks

1. Tag each serving endpoint. Endpoint tags reach the billing table, so cost splits by tag.
2. Use serverless usage policies. A policy tags serverless use with its team.
3. Create a budget. Admins get email when spend passes a threshold.
4. Track AI Gateway usage. Token use per endpoint goes into a system table.

Questions about Databricks