Rethinking the Model Provider Issue

[QAbot-zh]

Recent updates have created some user-unfriendly situations in terms of checking model names and providers. It's interesting and valuable to match a provider to a specific route and parameter format. However, I believe it's inappropriate to solely rely on model names to match providers. For example, if OpenAI releases a new chat model and if nextchat doesn't promptly include it in the DEFAULT_MODELS list, users won't be able to use this model. Furthermore, users would need to frequently update their nextchat version as new provider models become available, which isn't ideal. Perhaps the power to match providers with model names should be given to users by using CUSTOM_MODELS. Users could quickly adapt to new models by customizing model names. A possible definition format could be: gpt-4-turbo-preview|OpenAI, gemini-ultra|Google.