Unverified50% confidenceFactExact time
The researchers call the phenomenon where "the more tokens in the window, the harder it is for the model to reliably use any of them" "context rot."
1
Sources
50%
Confidence
Long-term
Relevance
10/25/2026
First Seen
Sources
Larger Context Windows Make AI Agents Less Reliable? 4 Failure Modes and How to Fix Them
bilibilicode炀10/24/2026
Related Entities
Related Claims
UnverifiedThe larger the context window, the more likely model performance is to degrade—the more information fed to the model, the worse it may perform.79% similarUnverified上下文窗口的大小不等于有效利用的能力，窗口越大指令衰减的潜在风险反而可能越突出72% similarUnverified模型规模越大，幻觉的表述往往越流畅、越具迷惑性，识别难度反而上升69% similarVerifiedSmaller models are more prone to embedding collapse than larger models, due to reasons including limited parameter capacity, a tendency toward optimization shortcuts, and amplified anisotropy issues.69% similarUnverifiedIn experiments, OpenClaw tended to lose direction and performed unstably when handling complex multi-step processes requiring contextual understanding.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1095306API
curl https://kongchang.com/api/v1/knowledge/claims/1095306MCP
get_claim(id=1095306)