Claude Token Counter, now with model comparisons

2026-04-19 · Source: Simon Willison's Weblog · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Intermediate, quick

Summary

The Claude Token Counter tool has been updated to allow direct comparison of token counts across different Claude models, specifically highlighting changes introduced with Claude Opus 4.7. Anthropic's Opus 4.7 utilizes an updated tokenizer, leading to an increase in token consumption for the same input, ranging from 1.0x to 1.35x for text. Testing revealed that the Opus 4.7 system prompt consumed 1.46x more tokens with the new tokenizer compared to Opus 4.6. Despite maintaining the same pricing ($5/million input, $25/million output), this token inflation effectively makes Opus 4.7 approximately 40% more expensive for text. For images, Opus 4.7 supports higher resolutions up to 2,576 pixels on the long edge, which can lead to a 3.01x increase in tokens for high-resolution images, though smaller images show negligible difference. A 30-page PDF showed a 1.08x token multiplier for Opus 4.7.

Key takeaway

For AI/ML teams evaluating Claude models, you should use the updated Claude Token Counter to precisely assess the cost implications of migrating to Opus 4.7. Be aware that while pricing per token remains constant, the new tokenizer in Opus 4.7 can increase effective costs by up to 40% for text and significantly more for high-resolution images. Factor these increased token counts into your budget and model selection decisions, especially for vision-intensive applications.

Key insights

Claude Opus 4.7's new tokenizer increases token counts and effective cost for text and high-resolution images.

Principles

• Tokenizer updates impact model cost.
• Higher resolution images consume more tokens.

Method

The updated Claude Token Counter tool enables direct comparison of token counts for text and images across different Claude models, particularly between Opus 4.7 and prior versions, to assess cost implications.

In practice

• Use the tool to compare token costs.
• Evaluate Opus 4.7's cost for high-res images.
• Check token inflation for specific text types.

Topics

• Claude Token Counter
• Claude Opus 4.7
• Tokenization Changes
• LLM Cost Analysis
• High-Resolution Image Support

Code references

• simonw/tools
• simonw/research

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Engineer, MLOps Engineer, Prompt Engineer

Related on AIssential

• Claude Opus 4.7 is AMAZING! Full Breakdown + Testing Results — Claude Opus 4.7 enhances image processing and design, but at a significantly higher operational cost.
• Your Claude Opus 4.7 Agent Ran Forty Steps. Here Is Why The Bill Tripled — Claude Opus 4.7's new tokenizer and other changes significantly increase token counts and costs for agentic workflows.
• Claude Message Limit Reached? Here’s What’s Actually Happening (And 2 Free Tools That Fix It) — Claude's free tier uses a dynamic compute quota, not message count, heavily impacted by conversation length.
• 5 Tips to Cut Claude Code Token Usage by 30% — Optimizing Claude Code usage through structured context and model selection significantly reduces token costs.
• Anthropic's Claude Opus 4.7 — Claude Opus 4.7 enhances agentic coding, computer vision, and knowledge work, but introduces tokenization and long-context performance tradeoffs.
• Claude Sonnet 4.6 in 7 mins! — Claude Sonnet 4.6 offers near-flagship performance at a lower nominal cost, but with higher token consumption and overeagerness.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Simon Willison's Weblog.