# [[Context rot]] ![[Context rot.svg]] [[Context]] rot is the phenomenon where the more information an LLM must recall, the less efficient it gets at doing so-- even before the [[Lost in the middle]] problem occurs. Context rot can be measured using the Needle in a Haystack experiment [^1] where a piece of information (the needle) is buried within a long document (of varying sizes) and the LLM is tasked with retrieving it. %% # Excalidraw Data ## Text Elements ## Drawing ```compressed-json N4KAkARALgngDgUwgLgAQQQDwMYEMA2AlgCYBOuA7hADTgQBuCpAzoQPYB2KqATLZMzYBXUtiRoIACyhQ4zZAHoFAc0JRJQgEYA6bGwC2CgF7N6hbEcK4OCtptbErHALRY8RMpWdx8Q1TdIEfARcZgRmBShcZQUebR44gAYaOiCEfQQOKGZuAG0AXX4IXDg4AGUoqHFUUDBIdXTqiCJlaRS6hkIECgAhXGwAa2VSYQ5iAGE2fDZSbggAYgAzZZX2 yGwRQKyASSr9CpGBhEnp2Yl5gEYEK6u1iA3SLahd9L7B4dGJqZm5qHIOZhwXBPO4PJ4vfQAMUI+HwFRgwTmgg8oM2mWeewObCOAHUSOpuHxwOs0TtMX9sQh4YiJMiSKjHuiIQAlYStDjhHJoC78EmMsnpADyQOwahg3AuiUSvPupIx6UhnCgkNw+hh4rQAFYZWCmXtFVkyoQjNUeNLibL+fL9AAVLBQACCLS4EmCiygDPB5OBjsebAokhCxG4HCE sJ1cohAFExg6/QGg3NgSMqBGrRC4ymbfAmiMhGNPXqFeQMqyaahQ+GLcxsCNYQANbgAZgALHEeC3EgBOLsAdibFxbLc1A5lNbr+AAmtwAGxdxLaAAcTZnbc1vcXFx42otRjYBm4tQ69AIQmqF2JAF80170qz88QOcwueg8wWZcMSEaTYTzR1P8QFQIHA3A7v+pAkAAsmwxAIDGuCaMEwZoIsBBhB+EFfKch7EpAPRTMhr7KJouAABQ8BcvbULwlH URRVGoAumoAJRrJAzIIMoYbAnMpDEWRPBNtKvBCXRomMdoLEQFeN7oliRzClA7AAiGYb4DKiwlggHFjEwhAcMoOF1JAmQIUh3B/GevLrEQIFoJZCDWRAHCqtUDlOcIUBEByFmkFZuEQPowJHKQAByrm+f5xmBcFTDwYhCCEQ5MkWnYABWCDYNkZQuXA0GwfF5koWhjkWv0SmMDa+74EZHSwIgSJpFlylsbKzBQAY2YNWglbqQF0yDIl3Cofg6EBZ pBhlE1SmcMNJVOfgoSOs1lXVapsIpR0jjMGZxxKvakGZEIc2jaVHSaPpjgGQACoEixMJkYhzKZCXBjKhDMD0l36cohVDfZflnZAH2QSQcBsPpUA5SUcA/X9SWAzKmhsJgU3BM1nD5U0ehZLg+nSWA150IsMLhIel4gJeQA== ``` %% [^1]: Hong, K., et al. (2025). *Context rot: How increasing input tokens impact LLM performance.* Retrieved in September 2026 from https://www.trychroma.com/research/context-rot .