Claude Opus 4.6 — The Smartest AI Model
Anthropic's most capable model. Industry-leading in agentic coding, research, finance, and tool use — with a 1M token context window and adaptive thinking.
Try Claude Opus 4.6
Enter a prompt below to experience Claude Opus 4.6's capabilities firsthand.
What Claude Opus 4.6 Can Do
Upgraded coding skills, deeper reasoning, and longer sustained agentic tasks — all in one model.
Agentic Coding
Plans more carefully, sustains tasks for longer, operates reliably in larger codebases, and catches its own mistakes with superior debugging skills.
Deep Research & Search
Outperforms all frontier models on BrowseComp for locating hard-to-find information. Leads on Humanity's Last Exam for complex multidisciplinary reasoning.
Finance & Knowledge Work
Outperforms GPT-5.2 by ~144 Elo points on GDPval-AA — an evaluation of economically valuable knowledge work in finance, legal, and other domains.
Benchmark Performance
State-of-the-art results across agentic coding, reasoning, search, and finance evaluations.
Terminal-Bench 2.0
Highest score on the agentic coding evaluation
Humanity's Last Exam
Leads all frontier models on complex multidisciplinary reasoning
BrowseComp
Best at locating hard-to-find information online
GDPval-AA
Outperforms GPT-5.2 by ~144 Elo points on knowledge work
MRCR v2 (1M)
8-needle 1M variant — vs Sonnet 4.5's 18.5%
SWE-bench Verified
Averaged over 25 trials on software engineering tasks
Key Features of Claude Opus 4.6
A comprehensive set of capabilities designed for developers, researchers, and knowledge workers.
Adaptive Thinking
Claude decides when deeper reasoning is helpful. Adjust effort levels (low, medium, high, max) to optimize for intelligence, speed, or cost.
1M Token Context (Beta)
First Opus-class model with 1M token context. Holds and tracks information over hundreds of thousands of tokens with less drift.
Context Compaction
Automatically summarizes and replaces older context when conversations approach the threshold, letting Claude perform longer tasks.
128K Output Tokens
Supports outputs of up to 128K tokens, enabling Claude to complete larger-output tasks without breaking them into multiple requests.
Enhanced Safety
Most comprehensive safety evaluations ever run. Low misaligned behavior rates and lowest over-refusal rate of any recent Claude model.
Agent Teams
Spin up multiple agents that work in parallel and coordinate autonomously — best for independent, read-heavy work like codebase reviews.
Frequently Asked Questions
Have more questions? Contact our support team.