# ARC-AGI-3 顶尖方案：LLM 引导符号世界模型合成

- 来源：François Chollet (@fchollet)
- 发布时间：2026-08-14 20:37
- AIHOT 分数：37
- AIHOT 链接：https://aihot.virxact.com/items/cmssy68420m0eroffag6yn1f0
- 原文链接：https://x.com/fchollet/status/2088243704603824311

## AI 摘要

François Chollet 指出，ARC-AGI-3 上所有顶尖方案均采用 LLM 引导的符号世界模型合成法，即通过编写可执行代码来编码对世界因果机制的理解。

## 正文

Jeremy's excellent work here is a great illustration of a very powerful type of approach: LLM-guided on-the-fly synthesis of a symbolic world model, i.e. making sense of the world by writing executable code that encodes your understanding of the causal mechanics of the world.

So far, all of the top-performing harnesses on ARC-AGI-3 use this style of approach. Which is also the approach we recommended when we initially released the benchmark (Jeremy would know this better than most, as a former Ndea member of technical staff).

I'm happy that ARC 3 has incentivized more research and more progress in this area.

### 引用推文

> Jeremy Berman：In one pass Opus wrote 269 programs (~12,700 lines). It built parsers for all 25 games, searching functions for 23, and game simulators for 9. It built a differ...