563
Dictionary

Chunk Size

The length of text segments stored in a vector index.

1 min readupdated 2026-07-04

/ quick answer

Chunk size trades off recall and precision. Small chunks (200-400 tokens) return tight, on-topic snippets but lose context; large chunks (800-1500) keep context but dilute similarity scores. The length of text segments stored in a vector index.

The length of text segments stored in a vector index. Chunk size trades off recall and precision. Small chunks (200-400 tokens) return tight, on-topic snippets but lose context; large chunks (800-1500) keep context but dilute similarity scores. In practice: A legal RAG uses 500-token chunks with 100-token overlap so clauses stay whole. This dictionary node is part of the Onexial knowledge graph and links to related concepts, workflows and tools below.
Definition
Chunk size trades off recall and precision. Small chunks (200-400 tokens) return tight, on-topic snippets but lose context; large chunks (800-1500) keep context but dilute similarity scores.
Example
A legal RAG uses 500-token chunks with 100-token overlap so clauses stay whole.
Related Tool Stacks
/ frequently asked

What is Chunk Size?

Chunk size trades off recall and precision. Small chunks (200-400 tokens) return tight, on-topic snippets but lose context; large chunks (800-1500) keep context but dilute similarity scores.

What is an example of Chunk Size?

A legal RAG uses 500-token chunks with 100-token overlap so clauses stay whole.

Why does Chunk Size matter for AI and automation?

The length of text segments stored in a vector index. It connects to the workflows, prompts and tool stacks linked on this page, so you can move from definition to execution without leaving Onexial.

/ continue exploring

Related concepts

The vocabulary this page depends on.

  • Retrieval

    Selecting the most relevant chunks for a query before generation.

  • RAG (Retrieval-Augmented Generation)

    Inject external knowledge into an LLM at query time.

  • Agentic RAG

    RAG where an agent decides what to retrieve, when, and from which source — instead of a single static query.

  • Embedding

    A numerical vector representation of text, image or audio that captures meaning for similarity search.

all dictionary

Related workflows

Turn this into a repeatable process.

all workflows

Related tool stacks

The tools that run it in production.

all tool stacks

Comparisons & alternatives

Pick between the options.

all comparisons