Startups: Companies that are in their initial stages of business and typically developing a business model and seeking financing.
Azure AI Foundry / Azure OpenAI usage that is covered by Microsoft for Startups sponsorship credits is still fully metered and billed; the sponsorship only offsets the charges on the invoice. The underlying consumption can be inspected through standard Azure cost tools, but it is important to use the right scope and views.
To address each point:
- Where to see the real pre-credit costs for Azure AI Foundry / OpenAI usage
Use Azure Cost Analysis scoped to the Azure OpenAI (or Azure AI Foundry) resource, resource group, or subscription:
- Sign in to the Azure portal.
- Open the specific Azure OpenAI / AI Foundry resource.
- Under Resource Management, select Cost analysis.
- By default, the view is scoped to that individual resource and shows the actual metered cost for that resource over time, before sponsorship credits are applied at the billing level.
The values shown here represent the “real” usage cost of the resource, regardless of whether sponsorship credits later reduce the invoice total. Sponsorship credits do not change or hide the metered resource cost; they are applied at billing settlement.
If multiple AI resources are used, repeat the above or change the scope in Cost Analysis to the relevant resource group or subscription to see aggregate costs.
- Reports/exports or detailed portal views per model or meter
Within Cost Analysis for the Azure OpenAI resource:
- In the Cost Analysis view, change Group by to Meter.
- Switch the chart type to Line (or use the table below the chart).
This shows which meters contribute to cost, such as different model series and token meters (for example, Text-Davinci tokens). This is the closest available breakdown to “per model/deployment” billing in the portal and lets identification of which model families are driving cost.
Key points from the cost-management guidance:
- Cost Analysis can show Azure OpenAI costs in graphs and tables by day, month, or year.
- Grouping by Meter reveals which model series and token meters make up the bulk of the cost.
- Scope can be adjusted to the resource, resource group, or subscription to understand the overall effect on costs.
If additional services are enabled (for example, Azure Monitor Logs, alerts), their costs appear under those services and at the subscription level, not under the Azure OpenAI resource itself.
- How to verify that AI usage is deducted from sponsorship balance
For Microsoft for Startups sponsorship credits, the key checks are:
- Confirm that the models in use are eligible for sponsorship coverage.
- Models sold directly by Azure are eligible and appear as standard Azure usage on the invoice. These are covered by sponsorship credits when used on a sponsored subscription.
- Models sold by partners and community (for example, Anthropic, Cohere select models, Hugging Face, Meta select models, Mistral AI select models) are not covered by sponsorship credits and are billed separately by the provider.
- On a sponsored Azure subscription, usage of eligible first-party models is automatically covered by Azure credits. No extra configuration is required.
From a billing perspective:
- Cost Analysis shows the full usage cost for Azure OpenAI resources.
- Sponsorship credits are then applied at the billing account level to offset those costs.
- Because credits offset the charges, the net amount due can remain unchanged between periods of low and high AI usage, even though the underlying resource cost (visible in Cost Analysis) increases.
To verify behavior end-to-end:
- In Cost Analysis for the Azure OpenAI resource, confirm that the time range covering heavy usage shows higher Azure OpenAI meter costs than a no-usage period.
- Ensure that the models used are in the “Direct from Azure” collection in Microsoft Foundry (these are billed directly by Azure and thus eligible for sponsorship credits).
- At the billing level (outside the scope of the provided documentation), sponsorship credits are applied against these metered costs; this is why the invoice total or “amount due” may not change even when Azure OpenAI resource costs increase.
If the Azure OpenAI resource’s Cost Analysis view does not show any increase in metered cost during heavy usage, despite confirmed calls, this would be abnormal and should be investigated with Azure Billing Support, as sponsorship credits do not suppress or zero out the metered cost itself.
References: