I have compiled some popular and well-designed open-source ChatGPT projects that can be deployed on Vercel, making it convenient for everyone to quickly build their own ChatGPT. Most of these projects support Vercel deployment and can be used with Vercel's free service. However, some projects include backend code and cannot be deployed on Vercel directly. For those, you will need to deploy them on your own server.
Recently, with the release of relevant APIs by OpenAI, more and more AI applications have emerged on the market. The chatpdf project caught my attention. How does it overcome the limitation of the maximum token of the API to read such long texts?