Anthropic Releases Claude Opus 4.6 With Million-Token Context and Autonomous Agent Teams.

Anthropic has released Claude Opus 4.6, a significant upgrade to its flagship AI model that introduces a million-token context window, improved reasoning capabilities, and new features across its product ecosystem. The update, announced on LinkedIn, is available today on claude.ai, the company's API, and major cloud platforms.

Opus 4.6 marks the first time an Opus-class model has featured a 1M token context window in beta, a substantial expansion from previous limits. This extended context allows the model to work with far larger codebases, documents, and datasets without losing track of earlier information. The model also demonstrates better planning, sustains agentic tasks for longer periods, and can catch and correct its own mistakes during execution.

The improved model applies its expanded capabilities to practical business tasks including financial analysis, research, and work with documents, spreadsheets, and presentations. Within Cowork, where Claude can multitask autonomously, Opus 4.6 can coordinate multiple workstreams independently.

Following its recent integration with Figma Make, the model is now powering more complex development workflows across Anthropic's product suite.

Claude Code now supports coordinating agent teams that work together autonomously. This allows developers to spin up multiple specialized agents that can divide complex coding tasks, share context, and collaborate without constant human oversight. The feature builds on the model's improved ability to maintain consistency across long-running operations and larger codebases.

For developers working in integrated development environments, this update complements existing tools like Cursor and Windsurf by offering native multi-agent coordination within Claude's ecosystem.

The API now includes adaptive thinking and effort controls, allowing developers to configure how deeply Claude should reason about a given task. This granular control helps balance response quality against computational cost for different use cases.

Context compaction addresses a common challenge in long-running tasks: hitting context limits mid-operation. The feature intelligently manages context to prevent workflows from stalling when they approach token boundaries, particularly valuable for extended research, analysis, or code generation tasks.

Claude in Excel now handles longer and more complex tasks with improved performance, making it more practical for financial modeling, data analysis, and other spreadsheet-heavy workflows that previously hit reliability or performance limits.

The company also introduced Claude in PowerPoint as a research preview. The integration allows users to build, edit, and iterate on presentations in real time, with Claude understanding existing deck structure and making contextual suggestions. This positions Claude alongside tools like Canva and Figma in the presentation creation space, though with a focus on AI-driven iteration rather than design-first templates.

Opus 4.6 is available immediately across all major deployment options: claude.ai for direct access, the Anthropic API for custom integrations, and major cloud platforms including AWS, Google Cloud, and Azure. Anthropic has not announced pricing changes with this release, though the expanded context window and improved capabilities may affect usage costs for API customers working with large documents or extended reasoning tasks.

The company recently announced it would offer Opus 4.6 to nonprofits at no additional cost, indicating the new model tier will carry different pricing than previous Opus versions for commercial users.

For small business owners and builders using AI coding tools, the combination of longer context, better error correction, and autonomous agent coordination represents a meaningful step forward in what's possible with AI-assisted development and business workflows.