Introducing Claude Opus 4.7

Anthropic has released Claude Opus 4.7 as the newest generally available version of its flagship Opus line, pitching it as a meaningful step forward for difficult engineering work, long-running tasks, multimodal understanding, and instruction fidelity. In the company’s launch post, Anthropic says Opus 4.7 improves materially on Opus 4.6 in advanced software engineering, especially on harder tasks that previously required tighter human supervision. The company also says the model now handles higher-resolution images better and produces stronger design-oriented outputs such as interfaces, slides, and documents.

Much of Anthropic’s announcement is built around evaluator and customer testimony, and the through-line is consistency under real workloads. Partners including Replit, Cursor, Devin, Vercel, Notion, CodeRabbit, Harvey, and others are quoted describing improvements in autonomy, tool use, long-context reasoning, and coding accuracy. Anthropic says pricing remains unchanged from Opus 4.6 at $5 per million input tokens and $25 per million output tokens, and that Opus 4.7 is available across Claude products, the Anthropic API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry.

The release also matters because of safety positioning. Anthropic says Opus 4.7 is the first model on which it is testing new cyber safeguards ahead of any broader release of its more powerful Mythos-class systems. The company describes automatic detection and blocking for prohibited or high-risk cybersecurity requests, while inviting legitimate security professionals into a new Cyber Verification Program. So while Opus 4.7 is being sold as a stronger coding and agent model, it is also functioning as a real-world test bed for the safety controls Anthropic wants in place before it pushes more capable systems into broader use.

Source: Anthropic official announcement — https://www.anthropic.com/news/claude-opus-4-7