Anthropic has released Claude Opus 4.5, a new model available on its apps, API, and major cloud platforms. The update includes pricing at $5/$25 per million tokens and enhancements to the Claude Developer Platform, Claude Code, and consumer applications.
- Opus 4.5 achieves state-of-the-art results on real-world software engineering tests and internal benchmarks, outperforming Sonnet 4.5.
- The model demonstrates significant efficiency gains, including cutting token usage in half for some tasks and reducing tool calling errors by 50% to 75%.
- It excels at long-horizon autonomous tasks, handling complex workflows with fewer dead-ends and showing a 15% improvement on Terminal Bench over Sonnet 4.5.
- Early access customers report improved performance in coding, Excel automation, and multi-step reasoning, with some tasks completing in thirty minutes compared to two hours previously.
The release aims to make Opus-level capabilities more accessible for enterprise use while providing developers with greater control over cost and efficiency through features like the effort parameter.