Anthropic Releases Claude Opus 4.1, upgrade to Opus 4

August 5, 2025

Details

  • Achieved a record-setting 74.5% accuracy on the SWE-bench Verified, surpassing both the original Opus 4 and competing models such as GPT-4.1.
  • Enhanced agentic capabilities in research, data analysis, creative writing, and complex reasoning tasks.
  • Introduced as a seamless, drop-in upgrade—no API endpoint or pricing changes from Opus 4.
  • Implemented new safety measures by disallowing simultaneous use of temperature and top_p settings in the API.
  • Automatically rolled out to all existing Opus users as of August 5, 2025.

Impact

‍
Anthropic's Claude Opus 4.1 reinforces the company's competitive edge, particularly in the developer and enterprise segments. The model’s enhanced coding accuracy sets a new standard, challenging rivals like OpenAI and Google to innovate further in agentic AI workflows. The ease of deployment encourages rapid adoption among enterprise customers, while the updated safety measures highlight Anthropic's ongoing commitment to predictability and responsible AI governance.

‍

Read more