Anthropic, a leading AI startup, today announced the release of Claude Opus 4.1, a new flagship model that delivers minor improvements over the Opus 4 model. Anthropic claims that Opus 4.1 performs better than Opus 4 on agentic tasks, real-world coding, and reasoning.
Anthropic mentioned in its announcement blog post that Opus 4.1 posted a state-of-the-art coding performance of 74.5% on SWE-bench Verified. On Agentic terminal coding, it scored 43.3%, up from Opus 4"s 39.2%. On AIME 2025, this updated model scored 78%, up from Opus 4"s 75.5%.
The Anthropic team wrote the following regarding Claude Opus 4.1 availability:
Claude Opus 4.1 is a drop-in replacement for Opus 4 that delivers superior performance and precision for real-world coding and agentic tasks. It handles complex, multi-step problems with more rigor and attention to detail.
Developers can now access Opus 4.1 for the same price via the Claude paid plan and in Claude Code. It is also available via Anthropic"s developer API, Amazon Bedrock, and Google Cloud"s Vertex AI. Anthropic recommends everyone switch to Opus 4.1 for better performance at the same price.
GitHub announced that Opus 4.1 is now available in GitHub Copilot for Enterprise and Pro+ plans. Developers can select this model in GitHub Copilot Chat on GitHub.com, Visual Studio Code, and GitHub Mobile.
It is important to note that Opus 4.1 will be available only in ask mode, and the older Opus 4 model will be deprecated in 15 days. GitHub mentioned that Opus 4.1 delivers notable performance gains in multi-file code refactoring when compared to the older Opus 4 model.
While the improvements in Opus 4.1 are modest, its immediate integration into major developer tools like GitHub Copilot offers tangible benefits to developers at no extra cost. Anthropic also announced its plan to release substantially larger improvements to its AI models in the coming weeks.