GPT-4o Series & the Microsoft Foundry Platform — Azure AI Foundry Architecture & Agent Service
GPT-4o Series & the Microsoft Foundry Platform
GPT-4.1 & GPT-4o Series on Microsoft Foundry
Microsoft made OpenAI's GPT-4.1 and GPT-4o series generally available in Microsoft Foundry (Azure AI Foundry), alongside general availability of the Asia-Pacific Data Zone and production-ready updates to Foundry Agent Service.
Four Models, Four Workloads
| Model | Best For |
|---|---|
| GPT-4.1 | Long-context engineering & agentic coding - 1M token context window, superior instruction following, and multi-step tool execution |
| GPT-4o | Omnimodal flagship - real-time text, vision, and audio processing across enterprise workloads |
| GPT-4.1 Mini | Balanced everyday work - matches or exceeds GPT-4o quality on coding and instruction tasks at 83% lower cost |
| GPT-4.1 Nano | Fastest and most affordable - high-volume, low-latency classification, routing, and extraction with 1M context |
Azure OpenAI Pricing (Standard Global, per million tokens)
| Model | Input | Cached Input (75% off) | Output | Context Window |
|---|---|---|---|---|
| GPT-4.1 | $2.00 | $0.50 | $8.00 | 1,047,576 tokens |
| GPT-4o | $2.50 | $1.25 (50% off) | $10.00 | 128,000 tokens |
| GPT-4.1 Mini | $0.40 | $0.10 | $1.60 | 1,047,576 tokens |
| GPT-4.1 Nano | $0.10 | $0.025 | $0.40 | 1,047,576 tokens |
The GPT-4.1 and GPT-4o families are available through Global Standard, Data Zones Standard, and Global Provisioned (PTU) deployment types across Foundry regions.
Asia-Pacific Data Zone & Anthropic Claude GA on Microsoft Foundry
Alongside GPT-6 Astra / Sol 6.1, GPT-5.x, and the GPT-4.1 / GPT-4o families, Anthropic Claude models (Claude Opus, Sonnet, and Haiku families) are Generally Available in Microsoft Foundry — offered as Azure-hosted deployments (Global and US Data Zone Standard) with native Microsoft Entra ID RBAC and Microsoft Azure Consumption Commitment (MACC) billing. Meanwhile, the Asia-Pacific (APAC) Data Zone lets APAC customers run frontier models while keeping data processing strictly within the APAC boundary alongside US and EU Data Zones.
Send an image URL along with structured prompt to GPT-4o on Foundry and stream response tokens.