Azure pricing page lists a new 'API Management (AI Gateway tier)' with 'Pricing details are coming soon'. Source
| Hard budgets (team / key / user) (scored) | partial Token quotas per counter key (e.g. subscription) per period, HTTP 403 when exceeded; quotas are in tokens, not currency |
|---|---|
| Rate & spend limits (scored) | yes Tokens per minute per counter key (HTTP 429) |
| Cost attribution / chargeback tags (scored) | yes Token metrics to Application Insights with up to 5 custom dimensions |
| FOCUS export (scored) | partial version: 1.2 (Azure bill) Azure Cost Management exports the Azure bill in FOCUS; per-consumer token data from APIM is not in that export |
| EU data residency option (scored) | yes Deployable in EU regions (e.g. West Europe) |
| Public pricing (scored) | yes |
| Open source | no Managed Azure service |
| Hosting | Both Managed tiers; self-hosted gateway on Premium (and Developer for evaluation) |
| Supported providers / models | unknown Azure OpenAI / Foundry models and OpenAI-compatible APIs; no count published |
| Caching | yes Semantic caching policies |
| Pricing model | Per unit-hour + per-request overage |
| Public price points | West Europe retail: Consumption first 1M calls free then $3.50/1M; Basic v2 ≈$150/mo (10M incl., $3/1M extra); Standard v2 ≈$700/mo (50M incl., $2.50/1M); new 'AI Gateway tier': price 'coming soon' |
Not verified yet: Supported providers / models. "unknown" means we could not confirm it in public sources. Vendors can send evidence to update a cell; see get verified. Verification never changes the scoring rules.