Token counter & cost estimator

It is an approximation (about 4 characters or 0.75 words per token for English). Each provider uses its own tokenizer, so exact counts differ by roughly 10–20%, more for code and non-English text.

Roughly 1,300 tokens of English prose.