Tokens & Context Windows
LLMs don't read words — they read tokens. Understanding tokenisation and context windows is the key to controlling cost, fitting your data, and choosing the right model.
Live Tokeniser
Type or paste any text to see how an LLM would break it into tokens. Each colour represents a different token type. Try the presets to see how tokenisation changes across languages, code, and prose.
Key insight: Tokenisation determines both cost and capacity. A 128K context window doesn't mean 128,000 words — it means 128,000 tokens, which is roughly 96,000 words in English. Code and non-Latin scripts produce far more tokens per word, so the same context window holds less of them. Understanding this ratio is essential for budgeting API costs and knowing what fits in a single conversation.
Context Window Comparator
See how the context windows of major models compare visually. The bar length shows relative capacity — hover for details. Use the calculator below to check if your content fits.
What fits? Calculator
Key insight: Context window size determines what you can fit into a single conversation — your prompt, any uploaded documents, AND the model's response all share the same window. A bigger window doesn't always mean better — smaller, focused inputs often produce better results than dumping everything in at once.