Partially Verified65% confidenceFactTime unknown
当前主流大语言模型（如GPT-4、Claude、Gemini）的上下文窗口是有限且临时的，每次对话结束后模型不会自动保留任何记忆
3
Sources
65%
Confidence
Long-term
Relevance
5/31/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedGPT-4、Claude 3.5 Sonnet、Gemini 1.5 Pro等新一代大语言模型具备多轮对话记忆、错误自动修正和跨文件上下文理解能力77% similarUnverifiedThe single-session context windows of the latest models such as Claude and GPT-4 have expanded to 128K or more tokens, but they still cannot retain memory across sessions77% similarUnverifiedEarly language models like GPT-3.5 only supported a 4K token context window73% similarUnverifiedModern large language models (such as the GPT series) almost universally adopt a decoder-only architecture, using masked self-attention to ensure that autoregressive generation does not leak future token information.70% similarUnverifiedCurrent mainstream large language models (such as GPT, Claude, etc.) adopt a stateless architecture, where each API call is independent and the model itself does not retain any user history on the server side.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/9005API
curl https://kongchang.com/api/v1/knowledge/claims/9005MCP
get_claim(id=9005)