TokenPad
RAG

Chunk Overlap Visualizer

See exactly what overlap duplicates, and what it costs you.

Your input

312 characters1 lines0 tokensor drop a file

Overlap VisualizerExact
0Token change
Token change0no change
Input tokens0what you pasted
Output tokens0what you would send
Token cost of this result
Output tokens0
As input$0.00
× 100K requests$0.00

Everything on this page runs in your browser. Nothing you paste is transmitted, because there is no server here to transmit it to.

Result

Why overlap exists

Meaning does not respect chunk boundaries. A sentence split across two chunks appears complete in neither, so neither embedding represents it and it becomes effectively unsearchable — a silent hole in the index.

Overlap guarantees that any given sentence appears whole in at least one chunk. The duplicated text marked in the output is what buys that guarantee.

What it costs, and where the ceiling is

Duplication raises the total token count of your index, which costs embedding money once and storage money permanently. Ten to twenty percent is the usual range and the cost is modest.

Above roughly twenty percent a different problem appears: near-duplicate chunks crowd each other out of the results, and a top-five becomes five views of the same paragraph. That is a retrieval quality loss, not just a storage cost.

Frequently asked questions

Is zero overlap ever right?
When chunks are already self-contained units — FAQ entries, product records, individual log lines. There is no boundary to protect, so overlap is pure waste.
Does overlap help with small chunks?
It matters more, proportionally, because a fixed overlap is a larger share of a small chunk. At very small sizes, increasing the chunk is usually a better fix than increasing the overlap.

More rag tools