konan 0.2.4

Blazingly fast text chunkers in Rust with pythonic bindings

konan

Like the paper angel of the Akatsuki, konan folds your documents into precise pieces — Rust chunkers behind a Python API.

---

Why konan?

• 🦀 Rust core — all chunking runs in native code, no Python-loop overhead
• ⚡ Multithreaded by default — chunk_many() fans out across all cores via rayon and releases the GIL
• 🌀 Real async — chunk_async / chunk_many_async are native async def, not thread-pool wrappers
• 🎯 Char-accurate offsets — text[chunk.start:chunk.end] == chunk.text, always (Python slicing semantics, emoji-safe)
• 🧠 Semantic chunking — splits on topic shifts using any OpenAI-compatible embeddings endpoint, or your own async embedder
• 🔌 Ports & adapters — the Embedder port is injectable; bring your own backend

Strategies

Chunker	Splits by	Best for
NaiveChunker	fixed word count	quick & dirty baselines
FixedSizeChunker	chars, sentence-aware, overlap	classic RAG pipelines
RecursiveChunker	separator hierarchy (\n\n → \n → … )	general text, LangChain-compatible
SentenceChunker	unicode sentence boundaries	prose, multilingual text
MarkdownChunker	document structure + heading breadcrumbs	docs, wikis, READMEs
TokenChunker	exact token counts (cl100k_base, o200k_base)	embedding-model token limits
SemanticChunker	embedding similarity drops	topic-coherent chunks

Installation

uv add konan        # or: pip install konan

Quickstart

from konan import RecursiveChunker

chunker = RecursiveChunker(chunk_size=1000, chunk_overlap=200)

chunks = chunker.chunk(open("moby_dick.txt").read())
print(chunks[0].text, chunks[0].start, chunks[0].end, chunks[0].hash)

Parallel — all cores, one call

# rayon work-stealing across every core, GIL released:
all_chunks = chunker.chunk_many(documents)

Async — real async def, no thread-pool wrappers

chunks = await chunker.chunk_async(text)
batches = await chunker.chunk_many_async(documents)

Markdown with breadcrumbs

from konan import MarkdownChunker

chunks = MarkdownChunker(chunk_size=800).chunk(readme_text)
# chunk text is prefixed with its heading trail: "# Guide > ## Install\n\n..."
# code fences are never split

Token-exact chunks

from konan import TokenChunker

chunker = TokenChunker(chunk_size=512, chunk_overlap=64, encoding="o200k_base")

Semantic chunking

Point it at any OpenAI-compatible /embeddings endpoint (OpenAI, vLLM, Ollama, LiteLLM, …):

from konan import OpenAIEmbedder, SemanticChunker

embedder = OpenAIEmbedder(
    base_url="https://api.openai.com/v1",
    model="text-embedding-3-small",
    api_key="sk-...",
    batch_size=128,    # texts per request
    timeout=30.0,      # request timeout, seconds
    max_retries=2,     # exponential backoff on 429/5xx/connect errors
    dimensions=512,    # optional: shorten text-embedding-3-* vectors
)
chunker = SemanticChunker(embedder=embedder, threshold=0.75)
chunks = await chunker.chunk_async(article)

Or inject your own embedder — any async callable works:

async def my_embedder(texts: list[str]) -> list[list[float]]:
    return await my_model.embed(texts)

chunker = SemanticChunker(embedder=my_embedder, percentile=95.0)
chunks = await chunker.chunk_async(article)   # async-only for Python embedders

Python embedders must return list[list[float]] — call .tolist() on numpy arrays. They are async-only: chunk()/chunk_many() raise a RuntimeError pointing you at the _async variants.

The Chunk object

chunk.text    # the chunk's text
chunk.start   # char offset into the source (Python slicing semantics)
chunk.end     # char offset, exclusive
chunk.index   # 0-based position
chunk.hash    # xxh3-64 content hash

Benchmarks

Benchmarked on Apple M3 Pro (arm64), Python 3.12.10. Decimal MB/s, median of 5 runs, measured from Python (the numbers you actually get). Reproduce both tables and plots with uv run --extra bench benchmarks/bench.py; Rust-level criterion benches: cargo bench -p konan-core.

vs other libraries (same 1 MB document, identical configs)

Strategy	Library	Throughput	Chunks
recursive	konan	282 MB/s	1298
recursive	semantic-text-splitter	129 MB/s	1368
recursive	chonkie	87 MB/s	1413
recursive	langchain-text-splitters	24 MB/s	1468
token	konan	68 MB/s	438
token	semchunk	24 MB/s	616
token	langchain-text-splitters	23 MB/s	438
token	chonkie	18 MB/s	438
token	semantic-text-splitter	4 MB/s	505
sentence	konan	305 MB/s	1042
sentence	chonkie	3 MB/s	2124
recursive (unicode)	semantic-text-splitter	167 MB/s	1150
recursive (unicode)	konan	166 MB/s	1092
recursive (unicode)	chonkie	57 MB/s	1171

Throughput per strategy (1 MB document)

Chunker	Config	Throughput	Chunks
NaiveChunker	200 words	553 MB/s	804
FixedSizeChunker	1000 chars, 200 overlap	1,602 MB/s	1255
RecursiveChunker	1000 chars, 200 overlap	282 MB/s	1298
SentenceChunker	1000 chars, 1 overlap	306 MB/s	1130
MarkdownChunker	1000 chars, 200 overlap	448 MB/s	1511
TokenChunker	512 tokens, 64 overlap (cl100k)	69 MB/s	438

Parallel scaling — rayon goes brrr

64 docs × 256 KB through RecursiveChunker:

Mode	Time	Throughput	Speedup
sequential chunk() loop	61 ms	267 MB/s	1.0×
chunk_many() (rayon, GIL released)	10 ms	1,568 MB/s	5.9×

(Pure-Rust criterion puts the same workload at ~6 GiB/s; the Python numbers include chunk-object conversion.)

Caveats, honestly: recursive/token use identical configs across libraries (1000 chars / 200 overlap; cl100k, 512 / 64 — note the identical token chunk counts for the libraries with the same windowing semantics). Sentence configs are not directly comparable (konan groups by chars, chonkie by tokens) — read those rows as per-library cost, not head-to-head. The recursive (unicode) rows run mixed German/CJK/Cyrillic/emoji prose, off konan's ASCII fast path. konan tokenizes with bpe-openai (tiktoken-equivalent output, much faster encoder).

Development

uv sync                       # set up the venv (builds the extension)
uv run maturin develop --uv   # rebuild after Rust changes
cargo test --workspace        # rust unit tests
uv run pytest -q              # python integration tests

The workspace is hexagonal: crates/konan-core is pure Rust (no PyO3) with Chunker and Embedder ports; crates/konan-py adapts it to Python.

License

MIT

---

_{Named after Konan of the Akatsuki — the only one who could fold paper into anything. 🗞️}