Anthropic Releases Claude Opus 4.1 With Major Coding and Reasoning Gains
Anthropic unveils Claude Opus 4.1, delivering significant upgrades in agentic tasks, real-world coding, and deep research capabilities. The updated model is available immediately across multiple platforms at the same price point as its predecessor.
Anthropic releases Claude Opus 4.1, an upgraded version of its flagship model that focuses heavily on improving agentic tasks, real-world coding, and complex reasoning. The new model achieves a state-of-the-art score of 74.5% on the SWE-bench Verified benchmark and offers enhanced skills for in-depth research and data analysis. Claude Opus 4.1 is available now for paid Claude users and in Claude Code, as well as through the Anthropic API, Amazon Bedrock, and Google Cloud's Vertex AI, all at the exact same pricing as Opus 4.
Industry partners report significant practical benefits from the upgrade. GitHub notes that Opus 4.1 outperforms its predecessor across most capabilities, with particularly strong gains in multi-file code refactoring. Meanwhile, Rakuten Group praises the model for its ability to pinpoint exact corrections within massive codebases without introducing unnecessary changes or bugs, making it an ideal tool for everyday debugging.
Windsurf measures the performance jump from Opus 4 to Opus 4.1 as a full standard deviation improvement on their junior developer benchmark, which matches the notable leap previously seen between Sonnet 3.7 and Sonnet 4. Anthropic strongly recommends that all developers upgrade to the new model immediately by using the claude-opus-4-1-20250805 API endpoint. The company also hints that even larger model improvements are scheduled for release in the coming weeks.