According to CoinMeta, Google has introduced two new robot models that enable robots to reason about each individual action, thus unlocking a wide range of tasks. For example, these models can make humanoid robots walk, squat, stretch, and manipulate objects to clean up messy rooms, and they can also cooperate with other robots to complete tasks more quickly. In addition, Google has also released two more robot models; the former is the most powerful embodied reasoning model, while the latter is the most efficient vision-language-action model. Both models can run locally on devices and quickly adapt to new robot entities.