A LITTLE ORDER, A LOT LESS WORK.
Text Chunker
Prepare text for retrieval workflows with overlap and explicit character positions.
01 — 18Start with data. Make room for work.
Batch processing available
INPUT
COMPARE WITH
Results will appear here
ONE OPERATION, MULTIPLE FILES
Do repeated work in one pass.
Files use the options above. Up to 10 files, 2 MiB each.
Files stay on your device. Download each result independently.
No files selected.
All processing happens in this browser.
How to use it
- Counts Unicode code points without splitting surrogate pairs. Choose chunk size and overlap.
- This is not a model tokenizer, semantic splitter or grapheme splitter. Limit: 2,000 chunks. Offsets are zero-based; end is exclusive.
Is data sent to a model?
No. These tools run locally in your browser without calling a model API.