Anthropic released Claude Opus 4.7 on April 16, 2026, bringing substantial improvements to vision capabilities and introducing a new effort level for fine-tuned reasoning control. The model is now generally available across all Claude products, API access, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry.
Claude Opus 4.7 Supports Images Up to 3.75 Megapixels
The most significant upgrade centers on vision capabilities. Claude Opus 4.7 now processes images up to 2,576 pixels on the long edge, equivalent to approximately 3.75 megapixels. This enhancement enables better understanding of technical diagrams, chemical structures, charts, and complex visual information that previously required simplification or multiple passes.
The improved multimodal understanding extends to professional applications including interface design, presentation slides, and document analysis where visual detail matters.
New 'xhigh' Effort Level Provides Finer Reasoning Control
Anthropic introduced a new 'xhigh' effort level that sits between 'high' and maximum reasoning. This addition gives developers more granular control over the trade-off between reasoning depth and response latency. The feature works alongside task budgets, now in public beta, which help guide token spend on specific tasks.
At higher effort levels, Claude Opus 4.7 produces more output tokens due to enhanced reasoning processes. An updated tokenizer also increases input token mapping by 1.0 to 1.35 times depending on content type.
Performance Improvements Across Software Engineering and Task Handling
Claude Opus 4.7 demonstrates notable advancement in software engineering tasks, with improved performance on difficult problems. The model shows enhanced instruction following, better attention to detail, and improved ability to verify outputs before reporting results.
The model handles complex, long-running tasks more effectively and includes improved file system-based memory across multi-session work. Anthropic describes the output as more tasteful and creative for professional applications.
New Safety Features and Unchanged Pricing
Anthropic added automated cyber safeguards that detect prohibited cybersecurity uses, alongside a Cyber Verification Program for legitimate security research. Despite the improvements, pricing remains unchanged at $5 per million input tokens and $25 per million output tokens.
While Claude Opus 4.7 shows stronger performance across multiple benchmark domains compared to Opus 4.6, Anthropic notes it remains less capable than Claude Mythos Preview. Community reaction on Hacker News was measured, with discussion focusing on incremental improvements rather than breakthrough capabilities.
Key Takeaways
- Claude Opus 4.7 supports images up to 2,576 pixels (3.75 megapixels), enabling better analysis of technical diagrams and complex visuals
- A new 'xhigh' effort level provides finer control between reasoning depth and response latency
- The model shows improved software engineering performance and better handling of complex, long-running tasks
- Pricing remains unchanged at $5 per million input tokens and $25 per million output tokens
- Generally available April 16, 2026 across Claude products, API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry