• AI第一次独立解开数学界的开放问题，Google DeepMind的Aletheia agent在Erdős猜想数据库中自主解决了4道未解题

• GUI世界模型不再用像素硬猜。Code2World把界面预测变成代码生成问题，8B模型媲美GPT-5

• 从视频里直接学控制策略，VideoWorld 2在真实手工任务上成功率提升70%，机器人操作同样有效

• 训练免费的推理框架，Chain of Mindset让模型在不同推理步骤自动切换"思维模式"，比最强baseline高近5%

阅读全文 →

• AI just solved real open problems in mathematics for the first time. Google DeepMind's Aletheia agent autonomously cracked 4 unsolved questions from the Erdos Conjectures database.

• GUI world models don't need pixel-level prediction. Code2World turns next-state forecasting into code generation — an 8B model rivals GPT-5.

• Learning control policies straight from video. VideoWorld 2 improves task success rate by up to 70% on real-world handcraft tasks, and the knowledge transfers to robot manipulation.

• A training-free reasoning framework that switches cognitive modes per step. Chain of Mindset beats the strongest baseline by nearly 5% across six benchmarks.

Read more →