Recent claims about massive context windows (e.g., 1 million tokens for Claude, 2 million for Grok) are misleading because LLMs are most effective at only approximately 20% token consumption; at 60% consumption, performance already shows steep declines.
factualpending
Evidence Quote
“LLMs are at their most effective, no matter how big this is, really only like 20% token consumption... as you start to use up more token usage, they get a little dumber and dumber to the point that not even 60% of the way, you have steep declines in performance.”
Created: 8/13/2026, 9:50:58 AM
My Notes
Loading notes...