AI cost calculator
Claude tokenizer change calculator
Anthropic says Claude 4.7 and later models use a new tokenizer that produces about 30% more tokens for the same text. Newer models are also cheaper per token, so the net effect on your bill depends on which model you move from and to.
Your inputs
Result
$1,290 to $1,079 a month (-16%)
Moving from Claude Sonnet 4.6 to Claude Sonnet 5.5 with 30% more tokens. The lower price per token more than covers the extra tokens.
- Today on Claude Sonnet 4.6
- $1,290 a month
- Tokens after the move
- 650M in, 52M out
- After on Claude Sonnet 5.5
- $1,079 a month
- Change
- -$211 a month-16.4%
- Extra tokens alone at today's prices
- +$387 a monthwhat the tokenizer change would cost with no price cut
Worked examples
Each row is the calculator run on the inputs shown, with every other input at its default.
| Move | Today | After | Change |
|---|---|---|---|
| Sonnet 4.6 to Sonnet 5.5 | $1,290 a month | $1,079 a month | -16.4% |
| Sonnet 4.6 to Opus 5.5 | $1,290 a month | $2,158 a month | +67.3% |
| Sonnet 4.6 to Opus 4.8 | $1,290 a month | $2,795 a month | +116.7% |
| Haiku 4.5 to Haiku 5.5 | $430 a month | $56 a month | -87.0% |
| Opus 4.6 to Opus 5.5 | $2,150 a month | $2,158 a month | +0.4% |
| Opus 4.6 to Fable 5.1 | $2,150 a month | $5,298 a month | +146.4% |
How it is calculated
The same text becomes more tokens on Claude 4.7 and later models, so the calculator raises your token counts before pricing them on the new model.
Anthropic's pricing docs say Claude 4.7 and later models use a newer tokenizer that produces approximately 30% more tokens for the same text, and that the exact increase depends on content and workload shape. Claude Sonnet 4.6 and earlier use the previous tokenizer.
The increase only applies when you move from an old-tokenizer model (Sonnet 4.6, 4.5, Opus 4.6, Haiku 4.5) to a new one. Haiku 5.5 charges a higher price for prompts over 100,000 tokens; this page uses its standard price. Cache writes are not counted.
Questions
Do newer Claude models use more tokens?
Yes. Anthropic says Claude 4.7 and later produce about 30% more tokens for the same text than Claude 4.6 and earlier, because of a new tokenizer.
Is Sonnet 5.5 cheaper than Sonnet 4.6?
Per token, yes: $2 / $10 against $3 / $15 per million. With 30% more tokens, the effective price is about $2.60 / $13, still about 13% below Sonnet 4.6.
Which models use the old tokenizer?
Claude Sonnet 4.6 and earlier, Opus 4.6 and earlier, and Haiku 4.5. Opus 4.7, Opus 4.8, Sonnet 5, Opus 5 and the 5.5 models use the new one.
How do I measure my own token increase?
Send a sample of your real prompts to the token counting endpoint for both models and compare. Code and non-English text can differ from the 30% average.
Sources
- Anthropic: Claude API pricing, checked 2026-10-29
Prices and plans change. This page was last checked on 2026-10-29. The data is open: llm-prices.json (CC BY 4.0).