With Claude Opus 4.6 now accessible in Microsoft Foundry, builders can delegate complicated duties finish‑to‑finish and belief the AI to execute independently in manufacturing.
At Microsoft we consider that intelligence and belief are the core necessities of agentic AI at scale. Constructed on Azure, Microsoft Foundry brings these capabilities collectively on a safe, scalable cloud basis for enterprise AI. Right this moment, with the launch of Claude Opus 4.6 in Microsoft Foundry, we deliver much more functionality to brokers that more and more study from and act on enterprise methods.
Claude Opus 4.6 brings Anthropic’s most superior reasoning capabilities into Microsoft Foundry, our interoperable platform the place intelligence and belief come collectively to allow autonomous work. In Foundry, Opus 4.6 can activate information from all over the place: leveraging Foundry IQ to entry knowledge from M365 Work IQ, Cloth IQ, and the online. The mannequin is finest utilized to complicated duties throughout coding, information work, and agent-driven workflows, supporting deeper reasoning whereas providing superior instruction following for reliability.
Builders can now delegate their most complicated work to AI methods for full-lifecycle growth from, necessities gathering to implementation and upkeep. Enterprise customers can generate paperwork, carry out analysis, and draft copy with skilled polish and area consciousness.
At Adobe, we’re constantly evaluating new AI capabilities that may assist us ship extra highly effective, accountable, and intuitive experiences for our prospects. We’ve been testing Claude fashions in Microsoft Foundry and are excited concerning the path of Anthropic’s mannequin roadmap. Foundry provides us a versatile, enterprise-ready atmosphere to discover frontier fashions whereas sustaining the belief, governance, and scale which are important for Adobe.
—Michael Marth, VP Engineering for Expertise Supervisor and LLM Optimizer, Adobe
Introducing Opus 4.6: Frontier intelligence, constructed for actual work
Claude Opus 4.6 is the newest model of Anthropic’s most clever mannequin, and is taken into account one of the best Opus mannequin for coding, enterprise brokers, {and professional} work. With a 1M token context window (beta) and 128K max output, Opus 4.6 is right for:
- Manufacturing code
- Subtle brokers
- Workplace duties
- Monetary evaluation
- Cybersecurity
- Laptop use
By combining Anthropic’s most superior mannequin with Foundry’s end-to-end tooling, prospects can transfer from experimentation to manufacturing quicker, with out stitching collectively infrastructure.
1. Autonomous coding at a brand new stage
Opus 4.6 handles massive codebases nicely and is especially efficient at long-running duties like refactoring, bug detection, and complicated implementations.
This enables senior engineers to delegate work that beforehand took days, overlaying every little thing from necessities gathering to implementation and upkeep, whereas staying targeted on critiques and decision-making.
With Foundry’s managed infrastructure and operational controls, groups can compress growth timelines from days into hours, whereas sustaining the rigor required for real-world manufacturing methods.
Macroscope has relied on Anthropic’s frontier fashions to push what’s doable in AI code evaluate, serving to us catch the bugs that matter most to prospects earlier than they ever attain manufacturing. We’re extremely excited to deepen this partnership by scaling on Azure infrastructure and delivering our product to groups all over the world.
—Kayvon Beykpour, CEO & Co-Founder, Macroscope
2. Higher information work throughout the enterprise
Opus 4.6 delivers a step-change in how enterprises method information work, which Anthropic defines throughout three pillars: search, analyze, and create. In Microsoft Foundry, these capabilities might be utilized on to enterprise knowledge, workflows, and agent-driven functions.
Opus 4.6 understands the conventions and norms {of professional} domains, producing paperwork, spreadsheets, and shows that look and browse like expert-created work. Mixed with Foundry’s governance and entry controls, this makes Opus 4.6 a terrific match for finance, authorized, and different precision-critical industries the place high quality, context, and compliance matter.
At Dentons, we’re scaling generative AI throughout drafting, evaluate, and analysis workflows throughout our international groups. Claude in Microsoft Foundry brings the frontier reasoning power we’d like for authorized work, backed by the governance and operational controls required in an enterprise atmosphere. Higher mannequin reasoning reduces rework and improves consistency, so our legal professionals can concentrate on larger worth judgment.
—Matej Jambrich, CTO, Dentons Europe
Excessive context monetary evaluation
Opus 4.6 excels at connecting insights throughout regulatory filings, market reviews, and inside enterprise knowledge, surfacing conclusions that will historically take analysts days to compile.
Its superior reasoning capabilities enable it to:
- Navigate nuanced monetary and regulatory contexts
- Generate compliance-sensitive outputs
- Preserve consistency and traceability throughout complicated analytical workflows
When deployed via Microsoft Foundry, these workflows profit from Azure’s safety, compliance, and auditability, serving to organizations apply frontier AI to high-stakes evaluation with confidence.
3. Advancing agentic and laptop use capabilities
In line with Anthropic, Opus 4.6 delivers main good points in laptop use, with robust efficiency on business benchmarks for visible understanding and multi-step navigation. Opus 4.6 units a brand new commonplace for laptop use. Claude can now function computer systems extra precisely, deal with extra complicated duties, and work throughout a number of functions seamlessly.
It could possibly work together with software program, navigate interfaces, full types, and transfer knowledge throughout functions. When deployed in Microsoft Foundry, these actions can run as safe, ruled brokers, enabling automation of workflows that span legacy methods, doc processing, and operational instruments.
Opus 4.6 can handle complicated, multi-tool workflows with much less oversight—a vital requirement for groups working AI methods at scale.
Anthropic on Azure offers Momentic with the reliability ensures wanted to course of thousands and thousands of tokens per hour on state-of-the-art fashions like Opus 4.5. Azure’s platform works seamlessly with Anthropic’s SDK, even supporting beta options reminiscent of reasoning effort out of the field.
—Jeff Ann, CTO & Co-Founder, Momentic AI
4. Brokers, safety, and high-stakes reasoning
Opus 4.6 can also be finest fitted to agentic workflows, reliably orchestrating complicated duties throughout dozens of instruments. It could possibly proactively spin up sub-agents, parallelize work, and drive duties ahead with minimal oversight.
For safety workflows, Opus 4.6 delivers deep reasoning, enabling groups to establish delicate patterns and complicated assault vectors with excessive accuracy.
Anthropic is a trusted associate for governments and corporations alike. Their pace, accuracy, and toolkit are already serving to Everstar make quick, secure nuclear power deployments a actuality. I’m excited to see these capabilities built-in natively on Azure for safe deployments for our authorities and nuclear prospects.
—Kevin Kong, Founder & CEO, Everstar
New API capabilities co-launching with Opus 4.6
Alongside Opus 4.6, Anthropic is introducing new API capabilities, accessible via Microsoft Foundry—that give builders higher management, scalability, and effectivity:
- Adaptive pondering: Permits Claude to dynamically resolve when and the way a lot reasoning is required, optimizing efficiency and pace on easier duties, whereas permitting Claude to suppose more durable on complicated duties.
- Context Compaction (beta): Helps long-running conversations and agentic workflows by summarizing older context as token limits are reached.
- 1M Context Window (beta): Assist for as much as 1M tokens, with premium pricing utilized past 200K tokens.
- Max effort control: A brand new max effort stage joins excessive, medium, and low, providing finer management over token allocation throughout pondering, instruments, and output.
- 128K Output Tokens: Permits richer, extra complete outputs in a single response.
Claude Opus 4.6 in Microsoft Foundry
As AI methods transfer from help to autonomy, success will depend on greater than frontier intelligence, it requires intelligence that may be trusted to function in real-world environments. Claude Opus 4.6 brings superior reasoning and long-horizon execution, and Microsoft Foundry offers the system context the place that intelligence might be utilized responsibly, at scale.
Collectively, Claude in Foundry allows organizations to maneuver past remoted experiments and towards clever, agent-driven methods that ship actual enterprise affect—grounded in belief, governance, and operational rigor.
Opus 4.6 can also be accessible via Microsoft Copilot Studio, enabling organizations to shortly construct, orchestrate, and deploy superior AI brokers with out customized code.