Claude Opus 4.6 — Our smartest model, upgraded

Model Announcement

Introducing Claude Opus 4.6

We're upgrading our smartest model. Claude Opus 4.6 improves on its predecessor's coding skills, sustains agentic tasks for longer, operates more reliably in larger codebases, and catches its own mistakes with better code review and debugging.

And in a first for our Opus-class models, Opus 4.6 features a 1M token context window in beta.

No. 1

Terminal-Bench 2.0 — Agentic Coding

+190

Elo over Opus 4.5 on GDPval-AA

Token context window (beta)

Capabilities

State-of-the-art across the board

Advanced Agentic Coding

Plans more carefully, navigates large codebases with confidence, and catches its own mistakes through improved code review and debugging. Highest score on Terminal-Bench 2.0.

Expert-Level Reasoning

Leads all frontier models on Humanity's Last Exam, a complex multidisciplinary reasoning test. Thinks more deeply and revisits its reasoning before settling on an answer.

Real-World Knowledge Work

Outperforms the industry's next-best model by 144 Elo on GDPval-AA — an evaluation of economically valuable tasks in finance, legal, and other professional domains.

Deep Agentic Search

Best-in-industry score on BrowseComp, measuring a model's ability to locate hard-to-find information online. Applies improved abilities to research, financial analyses, and document creation.

Bar charts comparing Claude Opus 4.6 to other models on GDPval-AA knowledge work benchmark

Platform

New platform capabilities

Agent Teams — Assemble multiple agents in Claude Code to work on tasks together.

Compaction — Claude summarizes its own context to perform longer-running tasks without hitting limits.

Adaptive Thinking — The model picks up on contextual cues about how much extended thinking to use.

Effort Controls — New developer controls to balance intelligence, speed, and cost.

Claude in Office — Substantial upgrades to Claude in Excel, plus a new Claude in PowerPoint research preview.

What partners are saying

"Early testing shows Claude Opus 4.6 delivering on the complex, multi-step coding work developers face every day — especially agentic workflows that demand planning and tool calling. This starts unlocking long-horizon tasks at the frontier."

Mario Rodriguez

Chief Product Officer, GitHub

Available now

Claude Opus 4.6 is live on claude.ai, the API, and all major cloud platforms. Pricing remains $5 / $25 per million tokens.

Try Claude

Read the Post

You are receiving this email because you opted-in to receive updates from Anthropic.

Anthropic, 548 Market St, San Francisco, CA 94104

Unsubscribe