Vercel

Anthropic's Claude Opus 4.7 Now on Vercel AI Gateway


Executive Summary

Anthropic has released its Claude Opus 4.7 model, which is now available on the Vercel AI Gateway. This new model is specifically optimized for complex, long-running agentic tasks that require reliable execution and visual analysis. Key enhancements include improved memory for multi-turn recall, stronger tool-calling for image and chart analysis, and a new "task budgets" feature for developers to control token usage.

Key Takeaways

* Availability: Claude Opus 4.7 is accessible immediately on the Vercel AI Gateway by setting the model to `anthropic/claude-opus-4.7`.

* Agentic Task Optimization: The model is engineered for asynchronous, multi-step tasks, demonstrating strong performance in agentic execution.

* Visual Capabilities: Features high-resolution image support and enhanced programmatic tool-calling, enabling pixel-level data transcription from charts and figures.

* Improved Memory: Agents using the model exhibit more reliable recall and retain facts better across conversational turns without extra prompting.

* Task Budgets: A new `taskBudget` feature allows developers to set a total token limit for an agent's turn, which the model uses to prioritize its work.

* New Developer Options: Introduces an `'xhigh'` effort level parameter for performance tuning and now omits "thinking" content by default to streamline outputs.

Strategic Importance

This launch makes Vercel's AI Gateway more competitive by offering developers a state-of-the-art model for building sophisticated AI agents that can handle complex workflows involving both text and visual data.

Original article