Particle.news
Download on the App Store
11 ARTICLES
·
4d ago

Anthropic Launches Claude Opus 4.1 to Boost Coding Accuracy and Agentic Reasoning

It marks a step in Anthropic’s rapid refinement strategy, boosting coding performance without raising prices.

Anthropic’s Dario Amodei and Daniela Amodei. Image: Anthropic
Image
Image
Geometric profile of hand and head with overlapping abstract shapes and silhouettes

Overview

  • Claude Opus 4.1 raises software engineering accuracy to 74.5% on SWE-bench Verified, up from 72.5% in Opus 4 and 62.3% in Sonnet 3.7.
  • It enhances agentic reasoning and data analysis by improving in-depth research skills and detail tracking.
  • GitHub notes marked gains in multi-file code refactoring compared with Opus 4.
  • Rakuten Group finds that Opus 4.1 pinpoints exact corrections in large codebases without introducing extraneous changes.
  • The update is available now to paid Claude users via Anthropic’s API, Claude Code, Amazon Bedrock and Google Cloud Vertex AI at the same price as its predecessor, with larger upgrades slated for the coming weeks.