Verified70% confidenceFactExact time
The "Lost in the Middle" effect refers to the phenomenon where models pay less attention to content injected into the middle positions of the context than to content at the beginning and end.
4
Sources
70%
Confidence
Long-term
Relevance
7/8/2026
First Seen
Sources
拆解ChatGPT选源机制：网络流量分析揭示AI引用真相
hackernewshackernews7/6/2026
Related Claims
UnverifiedAn "lost in the middle" effect exists in ultra-long contexts, meaning the model has stronger memory for content at the beginning and end of a sequence while attention to middle passages decreases; this phenomenon was named by Stanford researchers.77% similarUnverifiedLost in the Middle 效应指模型对位于极长输入序列中间位置的信息关注度会系统性下降，在超过 20K token 时开始显著，100K token 级别尤为突出77% similarVerified研究表明当上下文过长时，模型对中间部分信息的注意力会显著下降，即"Lost in the Middle"现象76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/262408API
curl https://kongchang.com/api/v1/knowledge/claims/262408MCP
get_claim(id=262408)