cancel
Showing results for 
Search instead for 
Did you mean: 
Generative AI
Explore discussions on generative artificial intelligence techniques and applications within the Databricks Community. Share ideas, challenges, and breakthroughs in this cutting-edge field.
cancel
Showing results for 
Search instead for 
Did you mean: 

Genie - Cost monitoring and usage

Sanjeeb2024
Valued Contributor

Hi Team,

Genie is widely used within our department, and we would like to establish appropriate governance around its usage. Do we have any existing guidelines or recommended practices for this? In particular, we are looking for guidance on:

  1. Cost monitoring: How can we monitor Genie usage, including who is using it and the associated consumption or cost?
  2. Usage limits: Is it possible to set an upper limit or other controls to restrict Genie usage?Any relevant documentation, recommendations, or examples would be appreciated.

Regards - Sanjeeb

Sanjeeb Mohapatra
1 ACCEPTED SOLUTION

Accepted Solutions

Ashwin_DSA
Databricks Employee
Databricks Employee

Hey @Sanjeeb2024,

Good to note Genie usage in your org. Databricks does have solid governance tooling for Genie, and it's matured quite a bit since billing went live on July 8, 2026. Here is what you need to know.

The source of truth for Genie spend is the system.billing.usage system table. You can query it directly (filter on billing_origin_product = 'GENIE') and join it with system.billing.list_prices to get dollar amounts. The identity_metadata.run_as column tells you who is using it, and usage_metadata.genie.surface breaks it down by products like Genie Code, Genie One, or Genie Agents. Full details and example SQL are in the Monitor and understand your Genie cost docs.

If you would rather not write SQL from scratch, account admins can import a pre-built cost management dashboard straight from the Account Console (Usage tab → Setup dashboard). 

For activity-level monitoring (who is asking what, feedback ratings, error rates), the system.access.audit table has you covered. The Monitor Genie Agents usage with audit logs page has ready-made queries you can drop in and set alerts on.

Account and workspace admins can also get a consolidated spend overview through the Cost page in Governance Hub.

To your second query about usage limits, this is where Budgets come in. They have been GA since July 6, 2026. To create a Genie-specific budget, go to Account Console → Usage → Budgets, select Unity AI Gateway as the resource type, and add the tag databricks-product: genie (don't add other tags, as they'll prevent the budget from tracking Genie usage).

You get two layers of control. A shared threshold sets an overall spending pool for all users and workspaces in scope. Once it's exhausted, everyone is blocked. A per-user threshold caps each individual's monthly pay-as-you-go spend (on top of their free 150 DBU/month allowance, which budgets cannot remove). You can also add per-user overrides for power users or specific groups who need a higher limit. For each threshold, you choose whether to just send an alert or actually block usage when the limit is hit. Databricks recommends alerting at the shared level and blocking at the per-user level.

It's also worth knowing that budgets are approximate (near real-time, not exact), they track LLM token cost only (SQL warehouse compute for running Genie-generated queries is billed separately), and you currently can't scope a budget to a specific Genie product (Code vs. One vs. Agents). They all share the same tag. There is also a Budgets API if you want to manage this programmatically. The Genie cost and budgets FAQ covers the most common edge cases.

I would suggest you to start with monitoring before you enforce limits. Set up a budget with alerts only so you can understand your baseline usage patterns. Once you have a feel for what normal looks like, add per-user blocking thresholds at a level that makes sense for your team (many orgs start around $30–$100/month per user). Use per-user overrides to give heavier users more headroom without raising the ceiling for everyone. And if chargeback matters to your organisation, be aware that custom cost-allocation tags don't currently propagate to Genie LLM cost, so you will need an external mapping to attribute spend to teams or projects.

For the full picture, the Cost management tools on Databricks overview page ties all of these capabilities together in one place.

Hope this helps.

If this answer resolves your question, could you mark it as “Accept as Solution”? That helps other users quickly find the correct fix.

Regards,
Ashwin | Delivery Solution Architect @ Databricks
Helping you build and scale the Data Intelligence Platform.
***Opinions are my own***

View solution in original post

4 REPLIES 4

balajij8
Esteemed Contributor II

@Sanjeeb2024 

You can use built-in Genie usage governance.

Cost monitoring - You can use the system.billing.usage system table for Genie usage with billing_origin_product, exposing per-user identity, product surface and DBU consumption. You can use it with system.billing.list_prices to calculate compute dollar costs and filter on sku_name to separate free allowance from billed usage. Every user receives 150 free DBUs/month for Genie Code. Genie One/Agents are free for users through January 31, 2027. More details here

You can use below to see per-user monthly free vs billed DBU and set alerts if required

SELECT
  u.identity_metadata.run_as AS user,
  u.usage_metadata.genie.surface AS genie_surface,
  SUM(CASE WHEN u.sku_name = 'GENIE_FREE_USAGE' THEN u.usage_quantity ELSE 0 END) AS free_dbus,
  SUM(CASE WHEN u.sku_name != 'GENIE_FREE_USAGE' THEN u.usage_quantity ELSE 0 END) AS paid_dbus
FROM system.billing.usage u
WHERE u.billing_origin_product = 'GENIE'
  AND u.usage_date >= DATE_TRUNC('MONTH', CURRENT_DATE)
GROUP BY u.identity_metadata.run_as, u.usage_metadata.genie.surface
ORDER BY user, genie_surface

Usage Limits - Account admins can create budgets scoped to Genie at the account, workspace, user-group, or individual-user level. Each budget threshold can trigger either an alert (email notification with continued access) or block usage (user is locked out until the budget resets or an admin raises the limit). Per-user overrides let you grant higher thresholds to specific groups, and users always retain their free monthly allowance regardless of budget settings. More details here

davidwarner344
Contributor

You can use usage/audit logs to track Genie activity by user and monitor consumption or costs. For governance, define approved use cases and set usage limits or alerts where available to avoid unexpected spend.

Ashwin_DSA
Databricks Employee
Databricks Employee

Hey @Sanjeeb2024,

Good to note Genie usage in your org. Databricks does have solid governance tooling for Genie, and it's matured quite a bit since billing went live on July 8, 2026. Here is what you need to know.

The source of truth for Genie spend is the system.billing.usage system table. You can query it directly (filter on billing_origin_product = 'GENIE') and join it with system.billing.list_prices to get dollar amounts. The identity_metadata.run_as column tells you who is using it, and usage_metadata.genie.surface breaks it down by products like Genie Code, Genie One, or Genie Agents. Full details and example SQL are in the Monitor and understand your Genie cost docs.

If you would rather not write SQL from scratch, account admins can import a pre-built cost management dashboard straight from the Account Console (Usage tab → Setup dashboard). 

For activity-level monitoring (who is asking what, feedback ratings, error rates), the system.access.audit table has you covered. The Monitor Genie Agents usage with audit logs page has ready-made queries you can drop in and set alerts on.

Account and workspace admins can also get a consolidated spend overview through the Cost page in Governance Hub.

To your second query about usage limits, this is where Budgets come in. They have been GA since July 6, 2026. To create a Genie-specific budget, go to Account Console → Usage → Budgets, select Unity AI Gateway as the resource type, and add the tag databricks-product: genie (don't add other tags, as they'll prevent the budget from tracking Genie usage).

You get two layers of control. A shared threshold sets an overall spending pool for all users and workspaces in scope. Once it's exhausted, everyone is blocked. A per-user threshold caps each individual's monthly pay-as-you-go spend (on top of their free 150 DBU/month allowance, which budgets cannot remove). You can also add per-user overrides for power users or specific groups who need a higher limit. For each threshold, you choose whether to just send an alert or actually block usage when the limit is hit. Databricks recommends alerting at the shared level and blocking at the per-user level.

It's also worth knowing that budgets are approximate (near real-time, not exact), they track LLM token cost only (SQL warehouse compute for running Genie-generated queries is billed separately), and you currently can't scope a budget to a specific Genie product (Code vs. One vs. Agents). They all share the same tag. There is also a Budgets API if you want to manage this programmatically. The Genie cost and budgets FAQ covers the most common edge cases.

I would suggest you to start with monitoring before you enforce limits. Set up a budget with alerts only so you can understand your baseline usage patterns. Once you have a feel for what normal looks like, add per-user blocking thresholds at a level that makes sense for your team (many orgs start around $30–$100/month per user). Use per-user overrides to give heavier users more headroom without raising the ceiling for everyone. And if chargeback matters to your organisation, be aware that custom cost-allocation tags don't currently propagate to Genie LLM cost, so you will need an external mapping to attribute spend to teams or projects.

For the full picture, the Cost management tools on Databricks overview page ties all of these capabilities together in one place.

Hope this helps.

If this answer resolves your question, could you mark it as “Accept as Solution”? That helps other users quickly find the correct fix.

Regards,
Ashwin | Delivery Solution Architect @ Databricks
Helping you build and scale the Data Intelligence Platform.
***Opinions are my own***

Sanjeeb2024
Valued Contributor

Thank you so much details. These are very useful details. I used Genie code to create a notebook which provide all details on the Genie cost, using those details I created a dashboard for monitoring and it provided the require details as of now, slowly we will enhance this and make a robust one.

Sanjeeb Mohapatra