I want to see if there is the possibility of being able to customize the models, for example increasing the tokens.
If there is a possibility, how can I make this adjustment?
Could you clarify what you mean by “increasing the tokens”?
If you’re attempting to increase your rate limits that can be found here.
I have an application that works like a chat that when it receives questions it consults a PDF and answers what is in the PDF.
But I want you to query multiple PDFs and I need a model that can interpret many PDFs.
The model that may work for me right now is gpt-4-1106-preview but is there a way to create a model with enough tokens?
If I’m understanding your question correctly. No sadly you can’t customize OpenAI models to get bigger input context windows. “gpt-4-1106-preview” is probably the best model with the biggest token limit. There might be other models with bigger limits, but there is a concept of how models with huge context windows don’t contextualize the information in the middle as well as information at the start or the end. All of that to say, if you have multiple PDFs that you want your model to be aware of, try out the new Assistants feature (either on API or Playground). Also look into RAG, it is what OpenAI uses in Assistants to reference files while not cramming them into your input.