ARC Prize：ARC-AGI-2 高分模型在 ARC-AGI-3 也更高
On ARC-AGI-3, there is a similar relationship: models that score higher on ARC-AGI-2 tend to also score higher on ARC-AGI-3.
AI 导读
ARC Prize 指出，ARC-AGI-3 上存在类似关系：在 ARC-AGI-2 得分更高的模型，往往在 ARC-AGI-3 上也得分更高。一个明显例外是低推理档的 Gemini 4 Argon，它在 ARC-AGI-2 上仅得 42.1%，但在 provider adapter harness 下于 ARC-AGI-3 上达到 97.5%。
正文
On ARC-AGI-3, there is a similar relationship: models that score higher on ARC-AGI-2 tend to also score higher on ARC-AGI-3.
One notable outlier is Gemini 4 Argon at low reasoning, which scored only 42.1% on ARC-AGI-2 but 97.5% on ARC-AGI-3 with the provider adapter harness. https://t.co/JQzMGRjJoq