Let's talk about the significance of Astra again. A few days ago, I mentioned that the core of Astra is to break through capabilities, not just improve performance. What is a breakthrough in ability? It is the generalization of computer usage capabilities. Astra is the first flagship model that can deliver completed computer work, enter engine editing scenarios, run long-term tasks overnight, and push business automation accuracy above competitors' level with lower API spending. Performance improvement usually refers to doing better on existing tasks; A breakthrough in ability makes a batch of tasks that could not be assigned to the model (which cannot be done well) worth assigning to it; The difference between the desktop tool Astra and all previous computer systems is not that it clicks more accurately, but that the interface layer has switched: the model no longer relies on various software vendors to choose open API toolboxes, but directly reads pixels and outputs keyboard and mouse events. In principle, any object that humans can see and click on belongs to its operating interface. The graphical interface provides another possibility: as long as there is an authorized and operable interface, the model has the opportunity to continue tasks through observation and operation. It reduces the requirement of having to redo specialized integration for every additional application or new process encountered. Existing software, file formats, and professional work environments have the opportunity to be more fully utilized by general models. We have seen many cases on X these days: 1) Many people directly throw photos to Astra and let it build editable models from scratch in Blender; 2) Let Astra play games on her own; 3) Use Astra to create a 3D promotional video for Zillow properties; 4) Or create a human anatomy site (2000+detachable parts); Wait All of them are proofs of the widespread use of computer skills in large models. The market may already have sufficient expectations for "writing better code"; Completing a whole piece of work across software makes it easier for people to imagine new scope of requirements. More tasks have become achievable with AI, increasing the demand for commercially viable computing. From this perspective, Astra provides new evidence to strengthen the long-term demand expectation for AI infrastructure. This article is sponsored by @ bitget_zh, titled 'Bitget Buying US Stocks: Instant Entry, Smooth Trading'