GPT-4 model API is used, but the response is that of the GPT-3 model.

Kazuki N 20 Reputation points
2023-08-29T02:50:40.6866667+00:00

Hi, I have already applied for a GPT-4 API application, it was approved and I am getting the GPT-4 API access.

But it seems that GPT-3 response is returned when I used GPT-4 model API, is it correct behavior?

I have deployed the GPT-4 model and am trying it out in Azure OpenAI Studio.

However, I suspect that the GPT-3 model response results are being returned even though the GPT-4 model API is being used.

Because when I throw the question "Are you using GPT-4 model?", the answer is that it is not a GPT-4 model.

Here is screenshot of response.

api_sample

However, since usage is occuring, I think that the API of the GPT-4 model is correctly called.

usage_sample

Please give us correct way to call GPT-4 model API or please make sure this behavior is correct or not.

Thanks.

Azure OpenAI Service
Azure OpenAI Service
An Azure service that provides access to OpenAI’s GPT-3 models with enterprise capabilities.
4,098 questions
0 comments No comments
{count} votes

Accepted answer
  1. romungi-MSFT 48,911 Reputation points Microsoft Employee Moderator
    2023-08-29T08:17:24.5733333+00:00

    @Kazuki N This is actually an expected behavior since the model is performing next token prediction in response to your question. The model doesn't have any native ability to query what model version is currently being run to answer your question. Since you have already confirmed that your deployment page shows GPT4 base model, the deployment is running the correct version. Please see this page from FAQ which answers this question.

    Quoting the FAQ for reference here.

    The questions, "What model are you running?" or "What is the latest model from OpenAI?" produce similar quality results to asking the model what the weather will be today. It might return the correct result, but purely by chance. On its own, the model has no real-world information other than what was part of its training/training data. In the case of GPT-4, as of August 2023 the underlying training data goes only up to September 2021. GPT-4 was not released until March 2023, so barring OpenAI releasing a new version with updated training data, or a new version that is fine-tuned to answer those specific questions, it's expected behavior for GPT-4 to respond that GPT-3 is the latest model release from OpenAI.

    If this answers your query, do click Accept Answer and Yes for was this answer helpful. And, if you have any further query do let us know.

    2 people found this answer helpful.
    0 comments No comments

0 additional answers

Sort by: Most helpful

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.