Token Calculator
Estimate token counts from raw text and see what it costs on any model.
Pricing dataset updated Oct 12, 2026
Estimated tokens
22
- Characters
- 86
- Words
- 15
- Cost as input
- $0.0000
- Cost as output
- $0.0001
Character-based estimate, not a provider tokenizer. Counts can differ substantially for code, other languages, and structured text. Confirm billing with the provider's token-count endpoint or usage report.
How it works
Tokens are sub-word fragments. For ordinary English prose, one token is roughly four characters or about 0.75 words; code and non-Latin scripts use more tokens per character.
The tool divides the pasted text’s character count by four and rounds up. Selecting a model changes the price used, not the token-counting method. It does not run that model’s tokenizer.
Use this as an early budgeting estimate for text, not an exact count or a context-limit check. Code, URLs, JSON, emoji, and different writing systems can produce substantial differences. For a production request, use the provider’s token-counting tools and confirm with its usage report.
Worked example
Estimating a support-response draft
A 2,400-character response draft produces an estimate of 600 tokens. At an illustrative input rate of $1 per million, the pasted text costs $0.0006 as input, or $6 across 10,000 such requests. Instructions and conversation history add to that total. Treat the number as a rough estimate until a provider tokenizer counts the complete request.
Provider references: Gemini token counting · Claude token counting