AI Cost Analysis

1. Confirm scope, period, currency, service, and the requested attribution: service, account/resource, meter or product, deployment, project, model, or token usage.
2. Query ServiceName first, then narrow with separate bounded queries by ResourceId, MeterCategory, MeterSubCategory, and Meter. Reuse exact returned values as filters.
3. Treat ServiceName and ResourceId as direct billing attribution. Treat meter-to-model interpretation as indirect unless the returned meter unambiguously names the model and price shape.
4. Cost Management does not expose native deployment, project, request, input token, or output token dimensions. Do not invent them. Resource tags may help map a resource to an owner or project, but they do not prove that all cost belongs to that tag value.

Correlate Usage

5. If a supported metrics source returns requests, tokens, throughput units, or deployment usage for the same resource and period, report it separately from billing cost. Derive unit cost only when the numerator and denominator have matching scope, time grain, and completeness.
6. Keep shared account charges, provisioned throughput, fine-tuning, storage, content safety, and other meters separate. Do not allocate shared charges to one deployment without a supported allocation source.
7. Preserve currency and period. If rows reach top, increase it to at most 5000 or narrow scope and label the result incomplete.

Report

Present the measured cost by service, resource, and meter; the requested attribution level; direct versus inferred mappings; any matching usage metrics; unallocated charges; confidence; and evidence gaps.

For future spend or planned model pricing, hand off to cost-estimation. A dedicated Microsoft Foundry workflow should be added only when supported model, deployment, project, and account attribution is established.