I think if you just put ‘gpt-4’ it will use the latest model. Which will be represented in the response. You can specifically request to use older models if needed (for a period of time). .
@xinort so you mean that even if we just put the “gpt-4”, we get the responses from “gpt-4-0314” because the latest version is “gpt-4-0314” now. Do I understand your reply well?
I have this same issue, however openai has pushed updates to gpt-4 since 0314, at least according to their own service status updates, so the 0314 model is most likely not the latest version of gpt-4.
Is there anyone else who had this issue but has managed to resolve it, and if so how did you do it?
Did we ever get a verdict on this? I just got access and it is in fact giving me
“model”: “gpt-4-0314”,
“object”: “chat.completion”
That model is almost 3 months old. Anyone figure out how to use the current “GPT4”?
More capable than any GPT-3.5 model, able to do more complex tasks, and optimized for chat. Will be updated with our latest model iteration.
8,192 tokens
Up to Sep 2021
gpt-4-0314
Snapshot of gpt-4 from March 14th 2023. Unlike gpt-4, this model will not receive updates, and will be deprecated 3 months after a new version is released.
GPT-4-0314 is a snapshot and is NOT updated. GPT-4 is updated. How can it be the most recent if it is clearly not updated like GPT-4?
Do you have documentation that says otherwise? It would not list GPT4 as available if it was not available, that’s why I am confused. Also, it is coming up on 3 months, 0314 will be depreciated.
I have no other documentation, but I read that as:
gpt-4 is basically an alias for whatever the latest version is. When a new model is released, it “Will be updated with our latest model iteration”.
gpt-4-0314 is snapshot. It’s not deprecated after 3 months, it’s deprecated 3 months “after a new version is released.”
So if they release a new model, let’s say gpt-4-0603:
gpt-4 would immediately resolve to gpt-4-0603
gpt-4-0314 would at that point be deprecated, then removed after 3 months. This allows developers to stick with the old version while they test the new model and adjust their logic/prompts as needed.