Share via

Azure open ai chat with your data for images not working

Srikantan Sankaran 20 Reputation points Microsoft Employee
2024-05-08T08:24:42.7966667+00:00

I have created a search index on an image catalog located in azure blob storage.

I was using this index to connect to from chat playground. I have used the gpt4 turbo vision preview model along with the AI Vision extensions to showcase ‘shop the look’ scenarios.

today, I noticed that when I enter the user request, it always comes back with an error ‘value must be provided for the key’. I have used the oai.azure.com playground.

i also notice that from the playground in ai.azure.com, I am unable to configure this index itself in the playground, when using the wizard experience.

Azure AI Search
Azure AI Search

An Azure search service with built-in artificial intelligence capabilities that enrich information to help identify and explore relevant content at scale.

Azure OpenAI in Foundry Models

Answer accepted by question author

  1. SnehaAgrawal-MSFT 22,721 Reputation points Moderator
    2024-05-10T05:14:04.6333333+00:00

    @Srikantan Sankaran It's recommended to utilize GPT4-0409, renowned for its robust vision capabilities, while GPT4-Turbo with Vision is slated for retirement.

    Please refer to the documentation at https://learn.microsoft.com/en-us/azure/ai-services/openai/gpt-v-quickstart?tabs=image%2Ccommand-line&pivots=programming-language-studio#differences-from-gpt-4-vision-preview for more details.

    The GPT-4 turbo-2024-04-09 version of GPT-4 is accessible for both standard and provisioned deployments. However, the provisioned variant of this model currently does not support image or vision inference requests.

    Our team is diligently working on updating the documentation and devising an alternative approach.

    Regrettably, we cannot provide an ETA at this moment.

    Stay tuned to QuickStart doc-

    1 person found this answer helpful.

0 additional answers

Sort by: Most helpful

Your answer

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