Token limits eased: one user argues current AI models are already smart enough

alexcovo_eth · x · 2026-11-19

alexcovoeth notes Anthropic and OpenAI token usage limits have improved so much he rarely approaches caps anymore. He argues frontier labs may not need ever-smarter models—superintelligent systems will instead train specialized small models and agents to handle everyday tasks, with bigger leaps reserved for complex problems and new discoveries.

Original post →

More from AGI Musings

AGI Musings channel →