SaaS Integrations

CostBeacon integrations let you connect external SaaS providers so their cost and usage data appears alongside your Azure spend. Today, integrations focus on third-party AI providers — giving you a single pane of glass for all AI-related costs, regardless of where the models are hosted.

Integrations are configured at the tenant level, meaning all users within a tenant see the same connected providers and their associated spend data. Only tenant administrators can add, edit, or remove integrations.

To connect a new provider, navigate to Admin → Integrations and click Add Integration. Select the provider from the dropdown, then enter your API key in the provided field.

When you save the integration, CostBeacon encrypts the API key and stores it in Azure Key Vault. The key is never stored in the application database and is not visible in the UI after the initial save. If you need to rotate a key, delete the existing integration and create a new one with the updated credentials.

Each provider can only be connected once per tenant. If an integration for a provider already exists, the Add Integration button for that provider will be disabled.

CostBeacon currently supports the following third-party providers:

• OpenAI — Connects using an Organization API key. CostBeacon calls the /v1/organization/costs endpoint to retrieve per-model cost breakdowns including GPT-4, GPT-4o, DALL-E, and embedding models.
• Anthropic — Connects using an Admin API key. CostBeacon calls the /v1/usage endpoint to fetch token-level usage data, then calculates costs based on published per-token pricing for Claude model variants.

Additional providers may be added in future releases. If you have a provider you would like to see supported, reach out via the feedback form in the app.

After adding an integration, use the Test button on the integration card to verify that CostBeacon can successfully authenticate with the provider's API. The test performs a lightweight read-only API call to validate that the key is correct and has sufficient permissions.

If the test fails, you will see an error message indicating the cause — typically an invalid key, expired key, or missing permissions. Correct the issue by deleting and re-creating the integration with a valid key.

A successful test does not trigger a full data sync. Data syncs happen on a separate schedule (see below).

Once a provider is connected and tested, CostBeacon syncs cost and usage data on a 12-hour cycle. The sync runs as a background job in the CostBeacon workers service, so it does not affect app performance.

During each sync, CostBeacon pulls the latest cost data from the provider's API, normalizes it into a common schema, and stores it in the CostBeacon database. The data then appears on the AI Spend page alongside Azure OpenAI and GPU VM costs.

The first sync after adding a new integration may take a few minutes depending on the volume of historical data. Subsequent syncs are incremental and typically complete in seconds.