Claude Opus 4.7 Launched

Anthropic has released a major update to its Opus model, introducing Claude Opus 4.7. This comes shortly after the launch of Claude Mythos, a highly advanced model designed for cybersecurity tasks with restricted access due to its powerful capabilities. While Opus 4.7 is built on a similar architecture, the company has intentionally limited its potential to prevent misuse in cyberattacks.

The newly launched model is now widely available across all Claude platforms, including API access and major enterprise services such as Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. Anthropic has retained the same pricing as the previous version, with input tokens costing $5 per million and output tokens priced at $25 per million.

Claude Opus 4.7 brings significant improvements in multimodal and coding tasks. It can now process high-resolution images more effectively, supporting up to 2,576 pixels, making it better suited for analyzing charts, screenshots, and documents. The model also performs better in complex coding tasks, maintaining consistency and verifying its outputs. Despite outperforming competing models in benchmarks, it still remains less advanced than the restricted Claude Mythos Preview.