# Anthropic Launches Latest Flagship Large Model Claude Opus 4.7

- Category: Flash (快讯)
- Published: 2026-04-17T11:58:00.000Z
- Source date: 2026-04-17
- HTML: https://www.88z.ai/en/news/flash-20260417-fdf4a265d17776e0
- Markdown: https://www.88z.ai/en/news/flash-20260417-fdf4a265d17776e0.md
- Other language: https://www.88z.ai/news/flash-20260417-fdf4a265d17776e0.md

Anthropic has officially released Claude Opus 4.7, positioning it as the most powerful large model currently available for widespread use. This upgrade focuses on enhancing complex task execution, high-definition visual understanding, and the stability of long-chain workflows, achieving significant breakthroughs in multiple benchmark tests.

## At a glance

- Visual capabilities jumped from 54.5% to 98.5% in XBOW tests
- Significant improvements in coding and long-context processing

## The update

Anthropic has officially launched its latest flagship large model, Claude Opus 4.7, aiming to establish it as the most powerful AI system currently available for broad deployment. The core of this update lies in comprehensively enhancing the model's ability to handle complex tasks, while significantly optimizing the accuracy of high-definition visual understanding and the operational stability of long-chain workflows. In terms of specific performance, the model's visual capabilities have achieved a qualitative leap. In the XBOW test, its score surged from 54.5% to 98.5%, approaching a perfect score. Regarding programming capabilities, the model achieved an 80.5% result in the SWE-bench benchmark. Furthermore, performance in the BFS test for million-token long contexts has improved markedly, with accuracy rising from 41.2% to 58.6%, demonstrating stronger long-text processing abilities.

## Sources

- [新智元](https://mp.weixin.qq.com/s/9m9S9Y9q9Fx4gscqizVDmg) (2026-04-17)

This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.
