Know how much of the context window your text will use.
Paste a sample, get a rough token estimate, and see how much room remains for instructions and the model’s answer.
Paste text on the left to estimate tokens and remaining context.
How to interpret the estimate
This calculator uses a transparent four-characters-per-token approximation. Real token counts depend on the provider tokenizer, language, punctuation, code, and document structure. English prose may land near the estimate, while source code, tables, and languages with different character patterns can vary substantially.
Do not fill a context window to its published maximum. Reserve room for system instructions, conversation history, retrieved passages, tool results, and the model’s answer. For production sizing, run representative documents through the provider’s official tokenizer and test whether retrieval or chunking produces more reliable answers than sending the entire document.