whichdevtool?
Menu
StacksDirectory Pricing changes Glossary
VPS and cloud servers App hosting and PaaS Managed databases Transactional email APIs Email marketing and newsletters Domain registrars CDN, DNS and edge security Monitoring, uptime and error tracking Auth and identity Frontend and full-stack frameworks Backend frameworks UI component libraries AI model APIs File and object storage Payments and billing Background jobs, queues and workflows ORMs and database toolkits

Cloud AI model platform · by Microsoft

Microsoft Foundry (Azure OpenAI)

Azure's model platform: OpenAI GPT models as Azure OpenAI plus Claude and other models, billed to Azure with Global, Data Zone (EU/US) and Regional deployments.

Visit azure.microsoft.com ↗ Alternatives Paid EU region
Microsoft Foundry (Azure OpenAI) homepage
§1

Overview

Microsoft Foundry is Azure's platform for AI models; the Azure OpenAI service within it hosts OpenAI's models, currently the GPT-5.4 family alongside older GPT-5.x and GPT-4 models. Claude models and other vendors' models are also offered, with Claude billed through the Azure Marketplace at Anthropic's list prices.

The deployment type sets both the price and where data is processed. Global Standard is the cheapest and routes worldwide; Data Zone deployments keep processing inside the EU or US data zone for roughly 10% more (GPT-5.4 mini is $0.75 in / $4.50 out globally versus $0.825 / $4.95 in a data zone); Regional deployments pin to one of many Azure regions; Provisioned Throughput Units reserve capacity on monthly or annual terms. The Batch API costs half the Global Standard price.

There is no free model allowance; usage is billed to your Azure subscription. The main reasons to choose Foundry over OpenAI's own API are buying through an existing Azure subscription and contract, and EU data-zone processing.

§2

Pricing and plans

Plan Price What you get
GPT-5.4 mini (Global Standard) $0.75 in / $4.50 out per 1M Cached input $0.075
GPT-5.4 mini (Data Zone) $0.825 in / $4.95 out per 1M Processing kept in the EU or US data zone
GPT-5.4 nano (Global Standard) $0.20 in / $1.25 out per 1M
Batch 50% off Global Standard e.g. GPT-5.4 mini batch $0.375 / $2.25; completes within 24 hours
Priority processing About 2x GPT-5.4 mini priority $1.50 / $9 (Global)
Provisioned (PTU) Monthly or annual reservation Reserved throughput units
Claude in Foundry Anthropic list prices Billed via Azure Marketplace in Claude Consumption Units ($0.01 each)

Read 2026-09-29. The Azure OpenAI pricing page displayed placeholder prices, so the GPT-5.4 figures above come from Microsoft's public Azure Retail Prices API (service 'Foundry Models', region East US 2, meters effective from March and April 2026). Prices can vary by region and by customer agreement; sign in to the Azure pricing calculator for your offer. Claude models in Foundry are rated at Anthropic's standard prices and invoiced through the Azure Marketplace; the US Data Zone deployment type applies Anthropic's 1.1x US-inference multiplier. No free tier for Foundry model usage was found.

Source: azure.microsoft.com/en-us/pricing/details/cognitive-services/openai-service/ · prices as published, excluding tax.

§3

What it does

Deployment types
Global, Data Zone (EU or US), Regional and Provisioned deployments with different prices.
Batch API
50% discount on Global Standard pricing for jobs completed within 24 hours.
Cached input
Reduced price for cached input tokens, e.g. $0.075 per 1M on GPT-5.4 mini.
Provisioned Throughput Units
Reserve model throughput with monthly or annual reservations.
Claude in Foundry
Anthropic's Claude models billed through Azure Marketplace at Anthropic's standard rates.
§4

Strengths and limitations

Strengths

  • OpenAI models billed through an Azure subscription.
  • EU Data Zone processing for about a 10% premium.
  • Batch halves Global Standard prices.

Limitations

  • No free tier.
  • The public pricing page may show placeholders; exact prices often need the pricing calculator or retail price API.
  • Data Zone and Regional deployments cost more than Global.
  • Provisioned capacity requires reservations.
§5

Who it suits

Good for

  • Enterprises standardised on Azure
  • EU companies that need in-EU processing for OpenAI models
  • Teams wanting OpenAI and Claude on one cloud bill

Look elsewhere if

  • Individuals without an Azure subscription
  • Hobby projects that need free usage
§6

Alternatives to Microsoft Foundry (Azure OpenAI)

Tool From Free option Stages
Usage-based Paid
Cerebras Inference Inference provider (open-weight models)
Usage-based Trial
OpenAI API AI model API (model maker)
Usage-based Paid
Claude API AI model API (model maker)
Usage-based Trial
OpenRouter Model router (unified API)
— Free tier
Together AI Inference provider (open-weight models)
Usage-based Paid

All AI model APIs →

§7

Questions

Is Azure OpenAI more expensive than OpenAI's API?

For GPT-5.4 mini, Azure's Global Standard price in East US 2 ($0.75 in / $4.50 out per 1M) matched OpenAI's standard price on 2026-09-29; Data Zone deployments cost about 10% more.

Can Azure OpenAI keep data in the EU?

Yes. Data Zone deployments keep processing within the EU (or US) data zone at a higher per-token price than Global.

Does Azure OpenAI have a Batch discount?

Yes. The Batch API is 50% off Global Standard pricing for requests completed within 24 hours.

Sources read for this page