Introducing Kimi K2.6 in Microsoft Foundry

We’re excited to welcome Moonshot AI’s Kimi K2.6 to Microsoft Foundry expanding the platform’s growing catalog of open and frontier models designed for real-world, production-grade AI systems.

Kimi K2.6 represents a new class of agentic, multimodal models built for long-horizon reasoning, coding, and autonomous execution—bringing developers closer to fully self-directed AI systems that can plan, act, and deliver outcomes end-to-end.

Why Kimi K2.6 matters

According to Moonshot AI, K2.6 is a native multimodal agentic model that advances capabilities in long-horizon coding, autonomous execution, and multi-agent orchestration.

This means developers can go beyond prompts and build systems where AI:

• Plans and executes multi-step workflows
• Writes, debugs, and refactors large codebases
• Generates full applications—from UI to backend
• Orchestrates multiple sub-agents to solve complex problems

What differentiates Kimi K2.6 is its focus on agentic intelligence at scale.

Unlike traditional models optimized for single responses, K2.6 is designed to:

• Handle long-running tasks across hundreds of steps
• Coordinate parallel sub-agents (“agent swarms”)
• Combine reasoning with tool use and execution
• Deliver complete outputs—documents, apps, workflows—in a single run

This aligns with the broader industry shift toward AI agents that operate more like systems than tools.

Built for developers: Coding, reasoning, and beyond

Kimi K2.6 builds on the Kimi K2 family, which introduced large-scale Mixture-of-Experts (MoE) architectures with up to 1 trillion parameters, optimized for reasoning, coding, and agent workflows.

With K2.6, those capabilities are extended further:

• Deeper reasoning and planning for complex, multi-file coding tasks
• Improved agent orchestration, enabling cleaner task decomposition
• Stronger tool-use reliability across multi-step workflows
• Multimodal inputs, combining text and visual understanding

The result is a model that is particularly well-suited for:

• Developer copilots and coding agents
• Document and knowledge workflows
• Autonomous research and analysis pipelines
• End-to-end application generation

Open models meet enterprise-grade infrastructure

Kimi K2.6 is part of a growing trend toward open, high-performance models that give developers flexibility without sacrificing capability.

In Microsoft Foundry, you can combine this openness with enterprise-grade features:

• Unified API and SDKs across models
• Model evaluation and observability tools
• Built-in safety and governance controls
• Flexible deployment options (global, regional, data zones)
• Integration with agent frameworks and orchestration tools

This means you can experiment with Kimi K2.6 and seamlessly move to production—without re-architecting your stack.

Pricing 

Getting started

Kimi K2.6 is now available in Microsoft Foundry.

You can:

• Explore the model in the Foundry catalog
• Benchmark it against other models using built-in evaluations
• Integrate it into your applications using the Foundry SDK