Simon Willison has updated his Claude Token Counter tool to allow comparisons between different Anthropic Claude models, specifically focusing on the tokenizer changes in Claude Opus 4.7. This new version reveals that Opus 4.7 can use up to 1.46x more tokens for text and significantly more for high-resolution images compared to Opus 4.6, despite maintaining the same pricing. The tool also highlights Opus 4.7's enhanced image processing capabilities, accepting images with longer edges up to 2,576 pixels. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
RANK_REASON Blog post detailing a tool update and analysis of a specific model's tokenizer changes.
