Paste text or upload .txt, .md, .pdf, or .docx. See if your document fits in Claude (200K / 1M), GPT-5.5 (400K), Gemini 2.5 (2M), or Llama 3 (128K) — plus a chunking strategy if it doesn't.
Every LLM has a maximum number of tokens it can process in a single call — the context window. Send more, and the model either errors or silently drops content. As of 2026-04-24: Claude Sonnet/Opus support 200K standard with 1M-token tier on Sonnet; GPT-5.5 supports 400K; Gemini 2.5 Pro supports 1M with a 2M tier; Llama 3.1 caps at 128K. This tool estimates your text size and shows whether each model can hold your full doc plus headroom for the prompt and response.
If your doc doesn't fit, the recommended strategy is chunking: split into overlapping sections, retrieve only relevant chunks per query (RAG), or use a 1M+ window model. Larger windows are not free — per-call cost scales linearly with input tokens, so even when a 2M window fits a doc, you may want to chunk for cost reasons.
All processing happens client-side: pdf.js renders PDFs to text, mammoth.js extracts plain text from .docx. Your file never leaves your browser. For exact token counts, use each provider's official tokenizer; these are heuristic estimates within ±5% of actual.