Verified80% confidenceFactExact time
大语言模型的上下文窗口是指模型在单次交互中能处理的最大文本长度，通常以token数量衡量
9
Sources
80%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026
Sources
The Essential Difference Between Agent Skills and MCP: Process Layer vs. Capability Layer
bilibiliAgent实战开发6/25/2026
Related Claims
Unverified大语言模型处理长任务的能力受制于上下文窗口的大小与质量83% similarVerified上下文窗口是指模型在单次对话中能处理的最大文本量，早期模型仅有4K Token，现代模型已扩展至数十万Token78% similarVerifiedA context window refers to the maximum number of tokens a large language model can process in a single inference76% similarUnverified大型语言模型的上下文本质上是模型在单次推理中能够处理的全部信息总量，通常以 Token 数量衡量76% similarUnverified大语言模型的总结功能本质上是通过注意力机制识别文本关键语义节点，将长文档压缩为保留核心论点的短文本75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/40722API
curl https://kongchang.com/api/v1/knowledge/claims/40722MCP
get_claim(id=40722)