The model is now generally available across all Claude products, the company's API, Amazon Bedrock, Google Cloud's Vertex AI, and Microsoft Foundry. Pricing is unchanged from its predecessor, Opus 4.6, at $5 per million input tokens and $25 per million output tokens.
Anthropic said Opus 4.7 delivers notable improvements on advanced coding tasks, including the ability to handle complex, long-running assignments with greater consistency, more precise instruction-following, and built-in output verification before reporting results. The company added that the model performs better than Opus 4.6 across a range of benchmarks, though it remains less broadly capable than Claude Mythos Preview, Anthropic's most powerful model.
On the vision front, Opus 4.7 now supports images up to 2,576 pixels on the long edge, more than three times the resolution limit of prior Claude models, enabling more detailed processing of screenshots, technical diagrams, and other visual inputs.
Anthropic also introduced cybersecurity-specific safeguards with the release. The company, which announced Project Glasswing last week highlighting AI-related security risks, said Opus 4.7 is the first model to carry automated systems designed to detect and block prohibited or high-risk cybersecurity requests. Security professionals seeking to use the model for legitimate purposes such as penetration testing or vulnerability research may apply through Anthropic's new Cyber Verification Program.
Accompanying the model release, Anthropic unveiled a new xhigh effort level for finer control over reasoning intensity, a /ultrareview command in Claude Code for in-depth code review sessions, and a public beta of task budgets on the Claude Platform API.
Developers can access the model using the identifier claude-opus-4-7 via the Claude API.
I truly appreciate you spending your valuable time here. To help make this blog the best it can be, I would love your feedback on this post. Let me know in the comments: How could this article be better? Was it clear? Did it have the right amount of detail? Did you notice any errors?
If you found any of the articles helpful, please consider sharing it.