🪙
Tokenizer
Estimate token usage for large language models.
🔒 Estimated locally in your browser.
0
Est. tokens
0
Characters
0
Words
How the estimate works
Exact token counts depend on each model's tokenizer (BPE). This gives a fast, dependency-free estimate by blending characters-per-token and words-per-token heuristics (≈4 chars or ≈0.75 words per token for English). For non-Latin scripts the estimate uses a higher tokens-per-character ratio. Treat it as a close approximation, not the exact billed count.