---
format: "aidr-story-markdown/v1"
id: "9c82c07570c3e559975ffb882fab4ee4e9702040ac0447b9bbf37c9430b87e61"
canonical_url: "https://aidr.today/9c82c075?lang=en"
title: "\"Next-token predictor\" is the wrong mental model for LLMs"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-09-04T17:09:24.000Z"
category: "Research"
topics: ["llm","reasoning"]
source_urls: ["https://news.ycombinator.com/item?id=49567310"]
summary: "Stop Thinking of LLMs as Next-Token Predictors"
---

# "Next\-token predictor" is the wrong mental model for LLMs

> [Open the canonical story](<https://aidr.today/9c82c075?lang=en>)

**Published:** 2026-09-04T17:09:24.000Z
**Category:** Research
**Topics:** llm, reasoning

## Summary

Stop Thinking of LLMs as Next\-Token Predictors

## Sources

- [Discussion](<https://news.ycombinator.com/item?id=49567310>)