Skip to content
Text Chunker for RAG & LLMs
Split long text into chunks by characters, words or estimated tokens with overlap, ready for embeddings, vector databases and RAG pipelines.
🔒 Processed locally in your browser — we don’t upload your input.
How to use the Text Chunker
Retrieval-augmented generation (RAG) systems store documents as small chunks in a vector database, retrieve the most relevant ones for each question and feed them to a language model. Chunk size and overlap strongly influence answer quality: chunks that are too large dilute relevance, chunks that are too small lose context. This tool lets you experiment with both settings and see the result instantly.
Choose to split by characters, words or estimated tokens, set the chunk size and the overlap, and optionally ask the chunker to respect paragraph and sentence boundaries so pieces do not end mid-sentence. Each chunk is shown with its index and length. You can copy all chunks as plain text separated by markers, or export them as JSON or JSONL with id and text fields, which most embedding scripts, LangChain, LlamaIndex and vector databases can ingest directly.
A common starting point is 300–800 tokens per chunk with 10–20% overlap, then tune based on your retrieval results. The chunker also works for splitting long transcripts for summarisation or translation. Processing is done in your browser, so confidential knowledge-base content never leaves your computer.
Frequently asked questions
What chunk size should I use for RAG?
Start with 300–800 tokens and 10–20% overlap, then test retrieval quality.
Why add overlap?
Overlap keeps sentences that cross a boundary present in both chunks, so context is not lost.
Can I export to JSONL?
Yes, export as JSON or JSONL with id and text fields.
Does it create embeddings?
No, it only splits text. Use your embedding model on the exported chunks.
About this tool
Split long text into chunks by characters, words or estimated tokens with overlap, ready for embeddings, vector databases and RAG pipelines. People also look for: text chunker, rag chunking tool, split text for embeddings, chunk size overlap.
Related searches