← All free toolsFree · no signup
Token counter.
Paste any text. See approximate token counts and per-model cost. Useful for sanity- checking prompt size, context-window fit, and per-call budget.
0 words·0 characters·~0 tokens (estimate)
About this estimate
Real tokenizers differ per model family — Anthropic, OpenAI, and Google each use their own. This counter uses a heuristic (max of words × 1.3 and chars ÷ 4) that lands within ±15% for English prose. Code, multilingual text, and unusual symbols may diverge more.
For exact counts in production, run the actual tokenizer client- or server-side (tiktoken for OpenAI, @anthropic-ai/tokenizer for Claude, the Gemini SDK for Google). For sizing prompts and budgets, this estimate is plenty.