semantic-chunking and Normalized-Semantic-Chunker
These two tools are competitors, with jparkerweb/semantic-chunking being the more established and widely adopted library for semantically chunking documents, while smart-models/Normalized-Semantic-Chunker is a newer, less-used alternative that claims to be "cutting-edge."
About semantic-chunking
jparkerweb/semantic-chunking
🍱 semantic-chunking ⇢ semantically create chunks from large document for passing to LLM workflows
Performs semantic chunking by embedding sentences with ONNX models and grouping them based on cosine similarity scores, with configurable thresholds and optional chunk rebalancing. Supports multiple embedding models including quantized variants (q4, q8), and can return chunk embeddings for RAG workflows. Deployable as a Node.js library, microservice API, or Docker container with an included web UI for interactive configuration.
About Normalized-Semantic-Chunker
smart-models/Normalized-Semantic-Chunker
Cutting-edge tool that unlocks the full potential of semantic chunking
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work