Library / SDK
Retrieval & Memory
Chonkie
1Lightweight 2fast 3text 4chunking 5for 6RAG
the six Ws · specification
W1
Who
RAG pipeline developers who need fast, flexible text splitting before embedding.
W2
What
Chonkie is a lightweight open-source chunking library offering token, sentence, semantic, recursive, and code-aware chunkers for RAG ingestion pipelines.
W3
Where
Self-hosted, installed as a Python (and TypeScript, via chonkiejs) package.
W4
When
Actively maintained with SIMD-accelerated chunking and coding-agent plugins added through 2026.
W5
Why
It exists to remove chunking as a bottleneck and quality risk in RAG ingestion pipelines.
W6
With
Integrates with tokenizer libraries and embedding models used downstream in RAG stacks.
W7
Watch
Open source, self-hosted Python/TypeScript library; no API key or external data transfer required.
alternatives
works with