Chunk Output: Count, Size & Distribution

AI Letters #18 — LLM Showdown #9: Chunking Strategies · Document: 1,972 chars · chunk_size=300, overlap=30

    The key finding: SynapseKit and LangChain produce identical output — 12 chunks averaging 163 chars. LlamaIndex produces 2 chunks averaging 986 chars. Same chunk_size=300 parameter, different interpretation: LangChain measures characters, LlamaIndex measures tokens (~4 chars/token).
  

Framework	Chunks produced	Avg size (chars)	Max size (chars)	chunk_size unit
SynapseKit	12	163	254	characters
LangChain	12	163	254	characters
LlamaIndex	2 ⚠️	986	1481	tokens

Chunk Count

Average Chunk Size (chars)

Lines of Code — Sentence-Aware Chunking

Chunk Size Visualization — Each Bar = One Chunk

SynapseKit — 12 chunks

LangChain — 12 chunks

LlamaIndex — 2 chunks (token-based: chunk_size=300 tokens ≈ 1,200 chars)