At Microsoft we believe that intelligence and trust are the core requirements of agentic AI at scale. Built on Azure, Microsoft Foundry brings these capabilities together on a secure, scalable cloud foundation for enterprise AI. Today, with the launch of Claude Opus 4.6 in Microsoft Foundry, we bring even more capability to agents that increasingly learn from and act on business systems.
Claude Opus 4.6 brings Anthropic’s most advanced reasoning capabilities into Microsoft Foundry, our interoperable platform where intelligence and trust come together to enable autonomous work. In Foundry, Opus 4.6 can activate knowledge from everywhere: leveraging Foundry IQ to access data from M365 Work IQ, Fabric IQ, and the web. The model is best applied to complex tasks across coding, knowledge work, and agent-driven workflows, supporting deeper reasoning while offering superior instruction following for reliability.
Developers can now delegate their most complex work to AI systems for full-lifecycle development from, requirements gathering to implementation and maintenance. Business users can generate documents, perform research, and draft copy with professional polish and domain awareness.
At Adobe, we’re continuously evaluating new AI capabilities that can help us deliver more powerful, responsible, and intuitive experiences for our customers. We’ve been testing Claude models in Microsoft Foundry and are excited about the direction of Anthropic’s model roadmap. Foundry gives us a flexible, enterprise-ready environment to explore frontier models while maintaining the trust, governance, and scale that are critical for Adobe.
—Michael Marth, VP Engineering for Experience Manager and LLM Optimizer, Adobe
Introducing Opus 4.6: Frontier intelligence, built for real work
Claude Opus 4.6 is the latest version of Anthropic’s most intelligent model, and is considered the best Opus model for coding, enterprise agents, and professional work. With a 1M token context window (beta) and 128K max output, Opus 4.6 is ideal for:
- Production code
- Sophisticated agents
- Office tasks
- Financial analysis
- Cybersecurity
- Computer use
By combining Anthropic’s most advanced model with Foundry’s end-to-end tooling, customers can move from experimentation to production faster, without stitching together infrastructure.
1. Autonomous coding at a new level
Opus 4.6 handles large codebases well and is particularly effective at long-running tasks like refactoring, bug detection, and complex implementations.
This allows senior engineers to delegate work that previously took days, covering everything from requirements gathering to implementation and maintenance, while staying focused on reviews and decision-making.
With Foundry’s managed infrastructure and operational controls, teams can compress development timelines from days into hours, while maintaining the rigor required for real-world production systems.
Macroscope has relied on Anthropic’s frontier models to push what’s possible in AI code review, helping us catch the bugs that matter most to customers before they ever reach production. We’re incredibly excited to deepen this partnership by scaling on Azure infrastructure and delivering our product to teams around the world.
—Kayvon Beykpour, CEO & Co-Founder, Macroscope
2. Better knowledge work across the enterprise
Opus 4.6 delivers a step-change in how enterprises approach knowledge work, which Anthropic defines across three pillars: search, analyze, and create. In Microsoft Foundry, these capabilities can be applied directly to enterprise data, workflows, and agent-driven applications.
Opus 4.6 understands the conventions and norms of professional domains, producing documents, spreadsheets, and presentations that look and read like expert-created work. Combined with Foundry’s governance and access controls, this makes Opus 4.6 a great fit for finance, legal, and other precision-critical industries where quality, context, and compliance matter.
At Dentons, we are scaling generative AI across drafting, review, and research workflows across our global teams. Claude in Microsoft Foundry brings the frontier reasoning strength we need for legal work, backed by the governance and operational controls required in an enterprise environment. Better model reasoning reduces rework and improves consistency, so our lawyers can focus on higher value judgment.
—Matej Jambrich, CTO, Dentons Europe
High context financial analysis
Opus 4.6 excels at connecting insights across regulatory filings, market reports, and internal enterprise data, surfacing conclusions that would traditionally take analysts days to compile.
Its advanced reasoning capabilities allow it to:
- Navigate nuanced financial and regulatory contexts
- Generate compliance-sensitive outputs
- Maintain consistency and traceability across complex analytical workflows
When deployed through Microsoft Foundry, these workflows benefit from Azure’s security, compliance, and auditability, helping organizations apply frontier AI to high-stakes analysis with confidence.
3. Advancing agentic and computer use capabilities
According to Anthropic, Opus 4.6 delivers major gains in computer use, with strong performance on industry benchmarks for visual understanding and multi-step navigation. Opus 4.6 sets a new standard for computer use. Claude can now operate computers more accurately, handle more complex tasks, and work across multiple applications seamlessly.
It can interact with software, navigate interfaces, complete forms, and move data across applications. When deployed in Microsoft Foundry, these actions can run as secure, governed agents, enabling automation of workflows that span legacy systems, document processing, and operational tools.
Opus 4.6 can manage complex, multi-tool workflows with less oversight—an essential requirement for teams operating AI systems at scale.
Anthropic on Azure provides Momentic with the reliability guarantees needed to process millions of tokens per hour on state-of-the-art models like Opus 4.5. Azure’s platform works seamlessly with Anthropic’s SDK, even supporting beta features such as reasoning effort out of the box.
—Jeff Ann, CTO & Co-Founder, Momentic AI
4. Agents, security, and high-stakes reasoning
Opus 4.6 is also best suited for agentic workflows, reliably orchestrating complex tasks across dozens of tools. It can proactively spin up sub-agents, parallelize work, and drive tasks forward with minimal oversight.
For security workflows, Opus 4.6 delivers deep reasoning, enabling teams to identify subtle patterns and complex attack vectors with high accuracy.
Anthropic is a trusted partner for governments and companies alike. Their speed, accuracy, and toolkit are already helping Everstar make fast, safe nuclear energy deployments a reality. I’m excited to see these capabilities integrated natively on Azure for secure deployments for our government and nuclear customers.
—Kevin Kong, Founder & CEO, Everstar
New API capabilities co-launching with Opus 4.6
Alongside Opus 4.6, Anthropic is introducing new API capabilities, available through Microsoft Foundry—that give developers greater control, scalability, and efficiency:
- Adaptive thinking: Allows Claude to dynamically decide when and how much reasoning is required, optimizing performance and speed on simpler tasks, while allowing Claude to think harder on complex tasks.
- Context Compaction (beta): Supports long-running conversations and agentic workflows by summarizing older context as token limits are reached.
- 1M Context Window (beta): Support for up to 1M tokens, with premium pricing applied beyond 200K tokens.
- Max effort control: A new max effort level joins high, medium, and low, offering finer control over token allocation across thinking, tools, and output.
- 128K Output Tokens: Enables richer, more comprehensive outputs in a single response.
Claude Opus 4.6 in Microsoft Foundry
As AI systems move from assistance to autonomy, success depends on more than frontier intelligence, it requires intelligence that can be trusted to operate in real-world environments. Claude Opus 4.6 brings advanced reasoning and long-horizon execution, and Microsoft Foundry provides the system context where that intelligence can be applied responsibly, at scale.
Together, Claude in Foundry enables organizations to move beyond isolated experiments and toward intelligent, agent-driven systems that deliver real business impact—grounded in trust, governance, and operational rigor.