Token Counter

Paste any text to estimate its token count, then see two things at a glance for every major model: what that text costs as input, and whether it fits the context window at all. Everything runs locally in your browser, so nothing is uploaded.

Token estimates use the standard English heuristic (~4 characters or ~¾ of a word per token, blended) and are typically within 15 to 20 percent of the real count. Every model tokenizes slightly differently, and code or non-English text runs denser. "Fits?" compares your text against each model's context window; "Used" is the share of that window your text would occupy, leaving the remainder for the conversation and the model's answer. For full workload costs including output tokens, use the cost calculator.