Meet the new
benchmark.

Anthropic · Now available

Claude Opus 4.1

A sharper upgrade to our most capable model — built for agentic tasks, real-world coding, and rigorous reasoning.

What's new
01
State-of-the-art coding
74.5% on SWE-bench Verified — leading real software engineering tasks.
02
Precise multi-file refactors
Surgical, accurate edits across large, complex codebases.
03
Deeper agentic reasoning
Stronger detail-tracking for long-horizon research & analysis.
By the numbers
74.5%
SWE-bench Verified
200K
Token context
4.1
Opus generation

Available now to paid Claude users, in the API, on Amazon Bedrock & Google Cloud Vertex AI.

Claude Opus 4.1

The most capable model, upgraded.

claude.ai
Anthropic · 2025