Anthropic spotted testing Claude Opus 4.1

Anthropic has been spotted working on the Claude Opus 4.1 model, hinting at next-level problem-solving capabilities for users. X user @legit_api has posted a screenshot, pointing out the model’s reference in the internal Claude code file. It has a description that says, “Try our latest model for more problem-solving power”.

This aligns with Anthropic’s recent claim of Claude Opus 4 achieving a 72.5% score on the SWE-bench, a heavy coding benchmark. The data suggests that Anthropic is paying heavy attention to this new model with next-generation upgrades. The two references in the code, including button_text-try and button_text-upgrade, show that the company could give this model for trials to early users.

In the past two years, Anthropic has steadily engaged in Claude’s development with faster and more efficient performance. The most recent Claude 4.0 includes both Opus 4 and Sonnet 4 have become famous for their coding capabilities.

Source – @legit_api

This is all thanks to the advanced reasoning and AI agents. The model is also profound in problem-solving, with tools like Cursor calling it state-of-the-art for coding.

On the SWE-bench Verified (a real-world software engineering benchmark), Opus 4 scored 72.5%, a significant jump from prior models. Sonnet 4 brings this frontier performance to all users, upgrading from Sonnet 3.7. A new feature, “thinking summaries,” uses a smaller model to condense lengthy reasoning (needed only 5% of the time), with a Developer Mode offering raw thought chains for advanced users.

For now, Anthropic hasn’t announced anything about the release of Claude Opus 4.1.