Microsoft Foundry Agent Service now supports a broader ecosystem of frontier and specialist models from providers including Anthropic, DeepSeek AI, Meta, Microsoft, and xAI, enabling enterprises to build flexible, enterprise-grade AI agents. Organizations want access to the best model for each scenario, without being locked into a single provider. This update delivers advanced reasoning, retrieval-augmented workflows, multimodal understanding, document intelligence, and high-performance agentic automation, all within Microsoft Foundry’s trusted, enterprise-ready framework.
Broader, Open Ecosystem of Models
Developers can now choose from an expanded set of cutting-edge models engineered for diverse strengths across reasoning depth, compute efficiency, context length, multimodality, and cost.
Newly supported models include:
Frontier reasoning
- DeepSeek-R1-0528 — advanced long-form and multi-step reasoning
- DeepSeek-V3-0324 — multimodal understanding across text and images
- DeepSeek-V3.1 — enhanced multimodal reasoning and grounded retrieval
- MAI-DS-R1 — deterministic, precision-focused reasoning
- grok-4 — frontier-scale reasoning for complex, multi-step problem solving
- grok-4-fast-reasoning — accelerated agentic reasoning optimized for workflow automation
Anthropic models
- Claude-Opus-4-1 — frontier reasoning for the most complex problem-solving
- Claude-Sonnet-4-5 — balanced performance for multimodal and agentic workflows
- Claude-Haiku-4-5 — lightweight, high-speed generation for interactive scenarios
General-purpose + enterprise orchestration
- Llama-4-Maverick-17B-128E-Instruct-FP8 — FP8-optimized model delivering fast, cost-efficient inference
- gpt-oss-120b — open-ecosystem model supporting transparency and reproducibility
xAI models
- grok-4-fast-non-reasoning — high-throughput, low-latency generation and system routing
- grok-3 — strong reasoning for complex, system-level workflows
- grok-3-mini — lightweight model optimized for interactive, high-volume use cases
This expanded catalog ensures organizations can choose the right model for every workload—without being tied to a single provider.
Built-in Capabilities for Enterprise Value
Foundry Agent Service turns models into production-ready agents with enterprise-grade features:
- Streaming responses for real-time, interactive engagement
- Flexible tool calling to connect with APIs, databases, and systems—supporting both structured queries and free-form inputs like SQL or scripts
- Structured outputs that integrate cleanly into downstream workflows
- Multimodal capabilities for analyzing documents, charts, visuals, and combined visual-textual reasoning
- File Search and Code Interpreter for grounded retrieval and safe, on-demand computation
- Model routing that automatically selects the best model for each task, balancing performance, accuracy, and cost
Enterprise Readiness at GA
All models run within Microsoft Foundry’s trusted, enterprise-grade environment:
- Governance and trust — RBAC, auditing, monitoring, and content safety controls
- Bring-your-own resources — private VNets, customer-managed Cosmos DB, and full data residency/retention controls
- Multi-agent workflows — orchestrate specialized agents across operations, supply chain, support, HR, finance, and more
- Open standards — Model Context Protocol (MCP) and Agent-to-Agent (A2A) communication for interoperability
- AgentOps — step-level tracing, evaluation, monitoring, and continuous improvement pipelines
These capabilities provide a seamless path from rapid prototyping to mission-critical deployment at scale.
Customer Quote
“Enterprises want choice and flexibility as they scale AI,” said John Tidwell, CTO of YoungWilliams. “With Microsoft Foundry’s expanded model ecosystem, we can pick the right model for each task—whether it’s reasoning, document intelligence, or automation—without being locked in. It helps us innovate faster while staying secure and compliant.”
Availability
- All newly added models are available via SDK, API, and the Foundry Agent Playground.
- Some models may require registration depending on region or compliance requirements.
Start building today
- Check new model availability by region in the model support documentation
- Get hands-on: Quickstart: Create and deploy a Foundry agent
Microsoft