Files are read and split on this device. Nothing is uploaded. Paste text or open a file, choose a chunk size, and get neat pieces you can copy or download as JSONL. It is free and private.
What chunking is for
Language models can only read so much at once, and search tools work best on small, focused passages. Splitting a long document into chunks lets you feed it to a model in parts, build a searchable knowledge base, or prepare data for retrieval. Doing it well means chunks that stand on their own and do not cut an idea in half.
How to split a document
- Paste your text into Text to split, or use Open a file. Try an example loads sample text.
- Choose whether to measure in Tokens or Characters.
- Set the Chunk size and the Overlap.
- Pick Natural breaks or Exact size.
- Optionally turn on New chunk at each Markdown heading.
- Check the numbers: chunks, average size, smallest and largest. Use Show more chunks to scroll the preview.
- Press Copy JSONL or Download JSONL.
Choosing a chunk size
For search and retrieval, chunks of about 200 to 800 tokens work well: big enough to hold a complete thought, small enough that a match points to the right place. Overlap of 10 to 20 percent keeps a sentence that falls on a boundary readable in both chunks. For summarising a long document in parts, use chunks close to what the model can read at once, minus room for its answer.
Natural breaks or exact size
Natural breaks keep paragraphs and sentences whole whenever they fit, so each chunk reads cleanly. Exact size fills every chunk as close to the limit as it can, which gives more even chunks at the cost of cutting mid-sentence. With Markdown headings turned on, each section starts a fresh chunk and records its heading path, such as Guide > Install.
Supported files
You can open text, Markdown, HTML, JSON, CSV and log files, and PDFs that contain a text layer. Page numbers are recorded for PDFs so you can trace a chunk back to its page.
Tips for good results
- Use the smallest size that still keeps a full idea together. Smaller chunks give sharper search results.
- Keep some overlap for prose, and little or none for tables or lists.
- Use the Markdown heading option for manuals, notes and documentation. The heading path travels with each chunk and gives it context.
- Look at the smallest and largest chunk in the summary. A very tiny chunk usually means a stray heading or a short line that is worth cleaning up first.
Who uses it
Students splitting a long reading to summarise it in parts, freelancers preparing client documents for search, and small teams building a help-desk knowledge base. To check how much a document will cost to process, try the AI cost estimator.
Questions
What is in the JSONL file?
One line per chunk with an id, the text, its number, the start and end position in the original text, its size in tokens or characters, the heading path for Markdown, the page numbers for a PDF, and the file name.
Which token counts are used?
With exact counts on, tokens are counted with the o200k tokenizer. Other models count a little differently, so leave some headroom if you're close to a hard limit.
Can it read scanned PDFs?
It reads the text layer of a PDF. A scanned PDF has no text layer; run it through the searchable PDF tool first, then open the result here.
How big a file can it handle?
Long books and reports are fine. Very large files take a few seconds, and the preview shows the first chunks with a button to show more; the download always has every chunk.
What is JSONL?
It is a plain text format with one JSON object per line. Most data tools and import scripts can read it, and it is easy to open in a text editor.