LLMs Aren't Fundamentally Discrete: Tokens Are Just Readouts of a Continuous Process
Sauers_ · x · 2026-11-11
Sauers pushes back on the claim that LLMs are fundamentally discontinuous because they emit discrete tokens. Tokens, he argues, are intermediate readouts from an underlying continuous process—and the text tokens capture only 0.001% of the information available internally. Equating discrete outputs with the model's entire "thought process" is a misconception.
More from Models
- Fable 5.2 cost per task doubles to $26.54, undercutting Anthropic's efficiency claims — haider1 · 2026-11-11
- Carbon tops Abundant Labs agent benchmark with 14/20 vs Opus 5.5 and GPT-6 — amankhan · 2026-11-11
- Fable 5.2 called an incremental, token-hungry update; Astra could retake lead — bindureddy · 2026-11-11
- sokobanteg v0.2 launches, benchmarking a dozen frontier models on Sokoban — banteg · 2026-11-11
- AssemblyAI launches Medical Mode with 20% fewer missed clinical terms — AssemblyAI · 2026-11-11
- Gemini 4 Carbon scores 99.1% on ARC-AGI-3 at $4.6K, beating Argon by 18 points — fchollet · 2026-11-11