Unverified50% confidenceSolutionExact time
When splitting text by token count, a chunk size of 200–500 tokens is typically set.
1
Sources
50%
Confidence
Long-term
Relevance
8/14/2026
First Seen
Sources
RAG实战：用检索增强生成打造百万级医疗问诊AI系统
Related Claims
Unverified一个精心编写的 500 token 规则文件往往比 3000 token 的冗长规则更有效70% similarUnverifiedIn the offline indexing phase of RAG, text chunks are typically split into 200–1000 tokens, and embedding vectors are typically 768 or 1536 dimensions.68% similarUnverified图片往往消耗约每张1000-2000 Token，而等效文字描述通常仅需100-300 Token66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/747966API
curl https://kongchang.com/api/v1/knowledge/claims/747966MCP
get_claim(id=747966)