Huawei and UCL's Memento 3 agent clears all 25 public ARC-AGI-3 games
Overview
Huawei Noah's Ark Lab and University College London released Memento 3, an agent that cleared all 25 public ARC-AGI-3 games and 183 levels using 7,518 actions, about 44% of the human total, according to Pandaily.
QbitAI reports the human baseline as 17,135 actions and a full RHAE score of 100.
Memento 3 runs on Claude Opus 5 at its highest reasoning setting, with no weight updates and no game-specific fine-tuning. The underlying model is not a Huawei model. In the official harness, the same base model averages about 40.7 RHAE.
The authors note that the public set does not test generalization to hidden games, that exploration is token-heavy, and that real-world settings remain untested.
Written by AI from the articles below · updated Oct 12, 5:00 AM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- PandailyNewsPickHuawei Noah's Ark Lab's Memento 3 clears all 25 ARC-AGI-3 public games with a frozen model
AIResearchers from Huawei Noah's Ark Lab and University College London released Memento 3, an agent that cleared all 25 public games in ARC-AGI-3 using 7,518 actions, about 44% of the human total. The agent runs on Claude Opus 5 without weight updates or game-specific fine-tuning, and the same base model averages about 40.7 RHAE in the official harness. The authors note that the public set does not test generalization to hidden games, and that exploration is token-heavy and real-world settings remain untested.
- QbitAINewsHuawei's Memento 3 agent clears 25 ARC-AGI-3 games with 56% fewer actions than humans
AIHuawei Noah's Ark Lab and UCL report that Memento 3, built on Claude Opus 5 with frozen weights, cleared all 25 public ARC-AGI-3 games and 183 levels. It used 7,518 actions against a human baseline of 17,135, earning a full RHAE score of 100. The system learns rules by writing a Markdown rulebook and an executable Python world model, and it later beat Atari Pong 21:0 in three games without calling the LLM.
Heat trend
Not enough continuous observations to show a trend yet.