Exploring the Model Catalog — The Model Catalog
Exploring the Model Catalog
Your AI Model Marketplace
The Foundry Model Catalog (Foundry Models) is one of the platform's most powerful features - a curated marketplace of 11,000+ AI models from frontier and open-weight providers, continuously updated with the latest releases.
Available Model Providers (October 2026)
| Provider | Key Models | Strengths |
|---|---|---|
| OpenAI (Azure OpenAI) | GPT-6 Astra, GPT-6.1 Sol, GPT-5.6 (Sol / Terra / Luna), GPT-4.1, GPT-4o, o3, o3-pro, o3-mini, o4-mini | Frontier reasoning, 1.05M context, OpenAI Responses API, zero customer data training, regional SLAs |
| Anthropic | Claude Opus 5.5 / 5, Claude Sonnet 5, Claude Haiku 4.5 | GA on Microsoft Foundry — Azure-hosted (Global & US Data Zone Standard), Entra ID RBAC, MACC-eligible |
| Microsoft (MAI) | MAI-Transcribe-1, MAI-Voice-1, MAI-Image-2, Phi-4, Phi-4-mini, Phi-4-reasoning, Phi-4-vision | First-party speech/vision, efficient on-device, extended Phi family |
| Microsoft Research | MagenticBrain, Fara1.5-9B | modern research models for specialized reasoning and efficiency |
| Meta | Llama 4 Maverick, Llama 4 Scout, Llama 3.3 70B | Natively multimodal MoE open-weight models, customizable |
| Mistral | Mistral Large, Ministral 3B | Efficient, multilingual |
| Alibaba | Qwen3 32B | Multilingual reasoning |
| xAI | Grok 4.7, Grok 4.6 | High-throughput frontier reasoning, real-time knowledge |
| DeepSeek & Fireworks AI | DeepSeek-R1, DeepSeek V4, DeepSeek V3.2, Kimi 2.6 | High-efficiency reasoning & ultra-fast open-weight serverless inference |
| Cohere | Command R+, Embed v3 | Enterprise RAG, embeddings |
Model Card Information
Every model has a Model Card containing benchmarks, license info, supported deployment types, pricing, and sample code.
GPT-6 Astra, GPT-6.1 Sol, GPT-5.6 & Frontier Reasoning on Microsoft Foundry
GPT-6 Astra (1.05M context), GPT-6.1 Sol, GPT-5.6 (Sol / Terra / Luna), and OpenAI o3 / o3-pro / o4-mini are Generally Available on Microsoft Foundry via Azure OpenAI Service. GPT-6 Astra provides autonomous multi-agent reasoning with up to 128K output tokens, while GPT-6.1 Sol delivers high-throughput reasoning (64K max output) at $3.50 / $14.00 per MTok with 90% prompt cache read discounts ($0.35/M cached). o3-pro and o3-mini provide configurable reasoning_effort (low, medium, high) for complex mathematical proofs, deep coding, and formal verification.
The Model Router
Foundry includes a Model Router (supporting Quality, Cost, and Balanced routing modes across 35+ models) that dynamically selects the optimal model per prompt in real time. Your application can automatically route complex reasoning to GPT-6 Astra, o3 / o4-mini, or Claude Opus, long-context tasks to GPT-6.1 Sol / GPT-4.1, and lightweight classification to Phi-4 — optimizing cost and latency without code changes.
azure-ai-inference SDK, Foundry Models standardize on the OpenAI v1-compatible API (/openai/v1/ Responses and Chat Completions via project.get_openai_client()), letting you swap OpenAI, Claude, Llama 4, DeepSeek, Mistral, and MAI models without rewriting client code.Search the Microsoft Foundry model catalog for frontier reasoning and vision models.