Text Chunker
Split text into overlapping chunks by tokens, characters, words, or paragraphs, then export as TXT, JSON, or JSONL.
Runs entirely in your browser. Nothing you type is uploaded.
About this tool
Paste text, choose a measurement unit, chunk size, overlap, and preferred boundaries, then inspect every generated chunk. Copy the result or export it as TXT, JSON, or JSONL; token counts are identified as estimates when no exact tokenizer is used.
Questions & answers
- What does chunk overlap do?
- Overlap repeats part of the previous chunk at the start of the next one, helping preserve context across boundaries. It must be smaller than the chunk size.
- Are token-sized chunks exact?
- Not necessarily. Unless a compatible tokenizer is selected, token sizes and per-chunk token counts are estimates and can differ from the target model.
- Which export format should I use?
- Use TXT for readable separated text, JSON for a structured array, and JSONL when a pipeline expects one chunk object per line.