Today, Vantage announces support for Baseten cost and usage data, giving customers the ability to track AI inference, training, and Model API consumption and costs directly in Vantage. Once connected, Vantage automatically ingests Baseten usage and cost data, allowing customers to monitor spend by product line (Dedicated Inference, Training, and Model APIs), model deployment, instance type, and model, set budgets, detect spend anomalies, and allocate Baseten costs alongside their broader cloud and AI infrastructure.

Baseten is a platform for deploying and running ML models in production, providing managed GPU infrastructure for inference and training. Enterprise customers often use it alongside major cloud providers to standardize model serving while keeping the rest of their infrastructure stack (data, networking, security, observability) on broader cloud providers such as AWS, Azure, and GCP. Previously, Vantage customers relied on manual exports from Baseten in order to bring costs into Vantage via Custom Providers.

Now, with Baseten cost support in Vantage, customers grant Vantage read access via a Baseten API key, and Vantage will automatically ingest their costs to display alongside their other infrastructure costs. Vantage retrieves detailed usage and cost data from the Baseten Billing Usage API, including costs broken down across Dedicated Inference (by model, instance type, and environment), Training (by job), and Model APIs (by model and model family, with input, output, and cached token counts). Customers can view Baseten spend alongside AWS, GCP, Azure, and other AI provider costs in Cost Reports, apply Virtual Tags to allocate inference and token costs to teams or products, and configure Budgets and Anomaly Detection on Baseten spend.

The Baseten integration is now available to all Vantage customers. To connect, go to the Integrations page in account settings and select Baseten. You can learn more via the Baseten section in Vantage's documentation.

Frequently Asked Questions

1. What is being launched today?

Vantage is launching native support for Baseten cost and usage data. Customers can now connect Baseten to Vantage, enabling automatic ingestion and visibility of AI inference, training, and Model API costs.

2. Who has access to this integration?

The Baseten integration is available to all Vantage customers across all subscription tiers.

3. How much does this integration cost?

There is no additional fee for using the integration. However, Baseten costs will be included in quota tier enforcement. If your Baseten costs push you over your current tier limit, you may be prompted to upgrade. See the Pricing page for details.

4. How does the integration work?

Vantage uses a Baseten API key to ingest usage and cost data from the Baseten Billing Usage Summary API (GET /v1/billing/usage_summary). After providing a Team API Key, Vantage begins ingesting cost and usage data broken down by Baseten product line, model deployment, instance type, and model, with daily granularity. For Model API consumption, Baseten only reports this usage in total price, and does not provide a cost breakdown. Vantage uses Baseten’s public pricing information and token split to perform further breakdown of tokens in order to be able to attribute Model API usage to input, output, and cached usage.

5. What permissions are required on the Baseten side?

You must be a Team Admin in order to create an API key for your organization. See the Baseten Teams documentation for more information.

6. What permissions are needed within Vantage?

You must have a Vantage Owner or Integration Owner role to add or remove the Baseten integration.

7. What permissions does Vantage have in my Baseten account?

In order to access the Baseten Billing Usage Summary API, the Team Key must be configured with Full Access. These are additional permissions beyond what Vantage uses; however, Vantage only reads data from the Billing Usage Summary API endpoint related to usage, and will never perform any other actions.

Unfortunately, Baseten does not currently support fine-grained permissions for Team API keys. We recommend customers submit a feature request to Baseten for improved permission scoping, and Vantage will adopt fine-grained permissions if this becomes available.

8. What dimensions can Baseten costs be filtered and grouped by?

Baseten reports support aggregating and filtering on the following dimensions:

• Organization (Name provided at time of integration)
• Service (Dedicated Inference, Training, Model APIs)
• Category (Model Name e.g., Qwen3 Next 80B A3B Instruct)
• Subcategory (Instance Type e.g., B200, H100 or Token Type e.g., Input Tokens, Output Tokens)
• Resource (Resource Name)
• Model Family (available via tags as baseten:model_family)
• Deployment (available via tags as baseten:deployment)
• Chain Name (available via tags as baseten:chain_name)
• Chain ID (available via tags as baseten:chain_id)
• Chain Deployment ID (available via tags as baseten:chain_deployment_id)
• Environment (available via tags as baseten:environment)
• Resource Kind (available via tags as baseten:resource_kind)
• Deletion Status (available via tags as baseten:is_deleted )

9. Can I view usage data for Baseten?

Yes. Usage data is available for Baseten. Dedicated Inference and Training report Compute Hours. Model APIs report in Tokens.

10. Are there Active Resources for Baseten?

No. Active Resources for Baseten are not available at launch.

11. How often does Baseten data refresh in the Vantage console?

Baseten data refreshes daily in the Vantage console.

12. What happens to deleted or torn-down deployments?

Historical cost attribution is preserved. The Baseten billing API continues to return deleted resources in the cost breakdown, so spend from a model deployment that has since been torn down still appears in your historical Baseten reports.

13. What happens if I remove the Baseten integration?

If you remove your Baseten integration from Vantage, all associated costs will be removed from the Vantage console.

14. How far back will the data go?

Vantage imports up to six months of historical Baseten cost data upon initial connection. Ongoing retention follows your Vantage data retention settings.

15. Can I have multiple Baseten account integrations?

Yes. You can connect multiple Baseten organizations as separate integrations in Vantage.

16. How are prepaid credits handled?

The Baseten billing API reports both gross subtotals and credits used. Vantage surfaces Baseten costs net of applied credits when your Cost Report is set to Credits enabled, so that reported spend reflects what you are actually billed. To see gross usage, set Credits to disabled.

17. Is tax included in costs from Baseten?

No, Baseten usage data does not return tax for your usage.