The research on AI is not yet very advanced, but it seems that OpenLora's approach is to serve models that are fine tuned through LoRA (Low Rank Adaptation) technology, improving concurrency, reducing costs, and meeting customized needs. I also have some ideas to see if they can be combined together. For example, implementing model assetization in Web3 scenarios. With the binding of on chain identity DID and LoRA model adapter, developers can encapsulate the model as NFT, and users have the right to call it once they hold it. This model not only stimulates the circulation of models, but may also give rise to an economic system where models are assets, which can help Web3 AI projects differentiate and compete. Of course, NFTs are just one possible force, and there are many possibilities in the future.