Tech & Energy
Context Window Calculator
Convert between tokens, words, pages, and characters — and see what fits in popular model context windows.
English words
—
- Pages (250 wpp)
- —
- Code lines
- —
- Characters
- —
FAQ
Reference context sizes? +
GPT-4o: 128k. Claude Sonnet 4.5: 200k (1M beta). Gemini 1.5 Pro: 2M. Llama 3.1: 128k. Most users only need 8-32k.
Why approximate? +
Tokenizers differ by model. BPE (GPT) is denser for English; SentencePiece (Llama) handles code better.