Anthropic released Claude Opus 4.7 on April 16, improving on Opus 4.6 across coding (13% gain on a 93-task benchmark), document reasoning (21% fewer errors), and vision (images up to 2,576px on the long edge). A new xhigh effort level gives developers finer control over the reasoning-vs-latency tradeoff, and a Cyber Verification Program unlocks expanded security research use cases. Pricing is unchanged at $5/$25 per million input/output tokens; the model is immediately available via the API, Amazon Bedrock, Google Vertex AI, and Microsoft Foundry.