UCL's Memento 3 Beats Human Efficiency by 56% on Every ARC-AGI-3 Puzzle

A frozen LLM agent learns by rewriting its own natural-language rulebook and compiling it into verified code, clearing every public ARC-AGI-3 game.

Source

AlphaSignal

Published

Oct 8, 2026, 12:06 PM

Author

AlphaSignal Newsroom

Read

1 min read

A frozen LLM agent learns by rewriting its own natural-language rulebook and compiling it into verified code, clearing every public ARC-AGI-3 game.

Reporting is indexed from AlphaSignal. Rights remain with the original publisher and cited sources.

Next reads