Unverified50% confidenceFactExact time

Even models that claim to support context windows of 128K or even million-level tokens frequently reveal weak mid-range retrieval in Needle-in-a-Haystack tests.

1

Sources

50%

Confidence

Medium-term (~90 days)

Relevance

9/1/2026

First Seen

Valid until: 11/30/2026

Sources

How to Find Research Gaps in ML/LLM/CV: A Systematic Methodology and Direction Guide

Related Entities

Needle-in-a-Haystackconcept

Related Claims

Unverified即使模型标称支持128K甚至1M token的上下文窗口，其有效利用能力往往远低于标称值，实际可靠工作范围可能只有标称值的几分之一66% similar Unverified该测试样本规模相对有限（5个模型、13个代码库），结论有待更大规模验证63% similar Unverified对抗样本攻击由Szegedy等人于2013年首次系统性揭示，只需叠加人眼不可见的微小扰动（幅度仅像素值1%~2%）即可让检测模型准确率从80%跌至接近随机水平60% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/838964

API

curl https://kongchang.com/api/v1/knowledge/claims/838964

MCP

get_claim(id=838964)