<p>Credit rating is a key element to reflect the level of credit risk in order to maintain the stability of financial markets. For enterprises, it supports the estimation of default risk to generate the credit rating and the associated funding costs. With the increase in credit risks over the past few years, for example, the falling of Silicon Valley Bank (SVB) and the Credit Suisse (CS), the demand for a more responsive and effective rating system is increasing. Although traditional models use structured data such as financial metrics, they are generally considered to have a time lag and incomprehensive, certain key elements that reflect the overall market environment, which might be generated from the unstructured data such as the changes in laws and policies as well as those that discussed in social media. Developing a more comprehensive and responsive credit rating system is a challenge to both practitioners and researchers. This study uses ChatGPT, a powerful LLM (Large Language Model), to collect a large set of multi-dimensional unstructured data and integrate such unstructured data with structured data to develop a comprehensive corporate rating model. Though the set of testing samples is small, the research results indicated that such an approach provides satisfactory accuracy and predictability. Even in the absence of some structured data from non-public financial reports, reasonable rating can be generated to provide the investors or business partners a reference, so this study shed the light to develop more suitable credit rating methodology for small and medium-sized enterprises, which were difficult to be rated with those traditional rating models.</p>