Document Text extraction model

Question

Document Text extraction model

Vusirikala, Aishwarya 0

I have created a model for "Document Text extraction" under Azure AI foundry. I would like to use the model that i have created in my python code. How do i use the model to extract the fields from the files?

Vikram Singh 2,575 Reputation points Microsoft Employee

2025-02-06T14:22:05.45+00:00
Hi Vusirikala, Aishwarya

Welcome to the Microsoft Q&A Forum! Thank you for posting your query here.

To utilize your custom "Document Text Extraction" model from Azure AI Foundry in your Python application. Here's how you can use the model to extract fields in your code:

Deploy the Model: First, ensure that your model is deployed in Azure AI Foundry. You can do this by navigating to your Azure AI Foundry project, selecting "Management center" from the left pane, and then selecting "Models + endpoints" under the Project section.

Get the REST Endpoint: Once your model is deployed, you will need the REST endpoint prediction URL. You can get this by selecting the model you want to deploy and noting the REST endpoint provided.

Use the Python SDK: You can use the Azure AI Document Intelligence SDK or the REST API to apply your custom model for inferencing.

Extract Fields: The response will contain the extracted fields from your document. You can parse this JSON response to get the specific fields you need.

For more detailed guidance, you can refer to the official Microsoft documentation: How to use Azure AI services in Azure AI Foundry portal

Additionally, the Azure AI Document Intelligence client library for Python provides detailed information on processing documents and extracting text using the SDK.

I hope this is helpful!

Thanks

Please do not forget to "Accept the answer” and “up-vote” wherever the information provided helps you, this can be beneficial to other community members.
Vikram Singh 2,575 Reputation points Microsoft Employee

2025-02-10T04:41:19.2533333+00:00

Hi Vusirikala, Aishwarya

Greetings.

Just following up to check if my suggestion helped. Please do not forget to "Accept the answer” and “up-vote” wherever the information provided helps you, this can be beneficial to other community members.

Thank you
Vusirikala, Aishwarya 0 Reputation points

2025-02-10T05:28:35.5966667+00:00

Thank you Vikram for your response. With your response, I am kinda getting some hope on how to proceed further.

below is what I see when i go to "Models + endpoints". I do not see the REST endpoint here. Can you guide me on how to get the REST endpoint?
Vusirikala, Aishwarya 0 Reputation points

2025-02-10T06:30:30.3233333+00:00
Hi Vikram,

Per Step 3 of your solution, i tried the Azure AI Document Intelligence SDK and below is what I get: Looks like the API version of the model created by AI Foundry is not being supported by the SDK. The API version created is 2024-07-31-preview. Any help on how to get around this issue?

File "C:\Users\VusirikaAi01\OneDrive - Hearst\FirstDataBank\HDA.venv\lib\site-packages\azure\ai\documentintelligence_operations_operations.py", line 819, in _analyze_document_initial

raise HttpResponseError(response=response, model=error)

azure.core.exceptions.HttpResponseError: (InvalidArgument) Invalid argument.

Code: InvalidArgument

Message: Invalid argument.

Inner error: {

"code": "InvalidParameter", "message": "The parameter CustomGenerative is invalid: Invalid build mode for this api-version."

}

PS C:\Users\VusirikaAi01\OneDrive - Hearst\FirstDataBank\HDA>
Vusirikala, Aishwarya 0 Reputation points

2025-02-10T06:35:49.5133333+00:00
Following up further:

Per Step #3, I went ahead and used the Azure AI Document Intelligence SDK and below is what i got: Looks like the API version of my model (2024-07-31-preview) is not being accepted by the SDK? How to get around with it?

File "C:\Users\VusirikaAi01\OneDrive - Hearst\FirstDataBank\HDA.venv\lib\site-packages\azure\ai\documentintelligence_operations_operations.py", line 819, in _analyze_document_initial

raise HttpResponseError(response=response, model=error)

azure.core.exceptions.HttpResponseError: (InvalidArgument) Invalid argument.

Code: InvalidArgument

Message: Invalid argument.

Inner error: {

"code": "InvalidParameter", "message": "The parameter CustomGenerative is invalid: Invalid build mode for this api-version."

}

PS C:\Users\VusirikaAi01\OneDrive - Hearst\FirstDataBank\HDA>
Vikram Singh 2,575 Reputation points Microsoft Employee

2025-02-10T07:45:23.5166667+00:00
Hello Vusirikala, Aishwarya,

Thank you for reaching out. It seems that the issue you're encountering is due to the API version 2024-07-31-preview not being supported by the Azure AI Document Intelligence SDK. Here are some steps to help you get around this issue:

Use a Supported API Version: The Azure AI Document Intelligence SDK may not yet support the 2024-07-31-preview API version. Try using a stable, generally available (GA) API version such as 2023-07-31 or 2023-10-3112. You can update your SDK client to target one of these supported API versions.

Update SDK Client Library: Ensure that you are using the latest version of the Azure AI Document Intelligence SDK. You can update the SDK client library by running the following command:
pip install --upgrade azure-ai-documentintelligence#This will ensure that you have the latest features and bug fixes

Modify API Request: Adjust your API request to use a supported API version. For example, you can modify the endpoint URL in your request to target the 2023-07-31 API version:
endpoint = "https://<your-endpoint-name>.cognitiveservices.azure.com/" api_version = "2023-07-31" url = f"{endpoint}/formrecognizer/documentModels/prebuilt-read:analyze?api-version={api_version}"

Check Parameter Validity: Ensure that all parameters in your API request are valid for the chosen API version. The error message indicates that the parameter CustomGenerative is invalid for the 2024-07-31-preview API version. Verify that you are using the correct parameters as per the API documentation3.

By following these steps, you should be able to resolve the InvalidArgument error and successfully use the Azure AI Document Intelligence SDK. If you need further assistance, feel free to ask!

For more detailed guidance, you can refer to the official Microsoft documentation:

Document Intelligence SDK target REST API 2023-07-31 (GA) latest.

Document Intelligence SDK target REST API v4.0

I hope this is helpful! Do not hesitate to let me know if you have any other questions.

Also, i think you resolve the REST endpoint you mentioned before.

Cheers
Vikram Singh 2,575 Reputation points Microsoft Employee

2025-02-12T11:59:46.99+00:00

Hello Vusirikala, Aishwarya,

Greetings.

Just following up to check if my suggestion helped. Please do not forget to "Accept the answer” and “up-vote” wherever the information provided helps you, this can be beneficial to other community members.

Thank you
Vikram Singh 2,575 Reputation points Microsoft Employee

2025-02-17T04:29:13.84+00:00

Hi Vusirikala, Aishwarya,

Greetings.

If the response helped, please do click Accept Answer and Yes for was this answer helpful.

Doing so would help other community members with similar issue identify the solution. I highly appreciate your contribution to the community.

Thank You.
Vikram Singh 2,575 Reputation points Microsoft Employee

2025-02-19T04:38:58.16+00:00

Hi Vusirikala, Aishwarya,

Greetings.

We haven’t heard from you on the last response and was just checking back to see if you got a chance to try above suggestions.

If the response helped, please don't forget to close up the thread here by upvoting and accept it as an answer.

Thank You.
Simon Ostermann 5 Reputation points

2025-04-16T14:53:20.74+00:00
I'm experiencing the exact same issue. The model created in AI Studio always defaults to API version 2024-07-31-preview, and I haven't found a way to change that. I'm using the Azure.AI.DocumentIntelligence .NET package, but I still encounter the same error.

Azure.RequestFailedException: Invalid argument. Status: 400 (Bad Request) ErrorCode: InvalidArgument Content: {"error":{"code":"InvalidArgument","message":"Invalid argument.","innererror":{"code":"InvalidParameter","message":"The parameter CustomGenerative is invalid: Invalid build mode for this api-version."}}}

Additionally, I can't change the API version of the SDK through DocumentIntelligenceClientOptions, since the enum only provides a single option, which is ServiceVersion.V2024_11_30.

1 answer

Your answer

Vikram Singh 2,575 Reputation points Microsoft Employee

2025-02-06T14:22:05.45+00:00

Hi Vusirikala, Aishwarya

Welcome to the Microsoft Q&A Forum! Thank you for posting your query here.

To utilize your custom "Document Text Extraction" model from Azure AI Foundry in your Python application. Here's how you can use the model to extract fields in your code:

Deploy the Model: First, ensure that your model is deployed in Azure AI Foundry. You can do this by navigating to your Azure AI Foundry project, selecting "Management center" from the left pane, and then selecting "Models + endpoints" under the Project section.

Get the REST Endpoint: Once your model is deployed, you will need the REST endpoint prediction URL. You can get this by selecting the model you want to deploy and noting the REST endpoint provided.

Use the Python SDK: You can use the Azure AI Document Intelligence SDK or the REST API to apply your custom model for inferencing.

Extract Fields: The response will contain the extracted fields from your document. You can parse this JSON response to get the specific fields you need.

For more detailed guidance, you can refer to the official Microsoft documentation: How to use Azure AI services in Azure AI Foundry portal

Additionally, the Azure AI Document Intelligence client library for Python provides detailed information on processing documents and extracting text using the SDK.

I hope this is helpful!

Thanks

Please do not forget to "Accept the answer” and “up-vote” wherever the information provided helps you, this can be beneficial to other community members.
Vikram Singh 2,575 Reputation points Microsoft Employee

2025-02-10T04:41:19.2533333+00:00

Hi Vusirikala, Aishwarya

Greetings.

Just following up to check if my suggestion helped. Please do not forget to "Accept the answer” and “up-vote” wherever the information provided helps you, this can be beneficial to other community members.

Thank you
Vusirikala, Aishwarya 0 Reputation points

2025-02-10T05:28:35.5966667+00:00

Thank you Vikram for your response. With your response, I am kinda getting some hope on how to proceed further.

below is what I see when i go to "Models + endpoints". I do not see the REST endpoint here. Can you guide me on how to get the REST endpoint?
Vusirikala, Aishwarya 0 Reputation points

2025-02-10T06:30:30.3233333+00:00

Hi Vikram,

Per Step 3 of your solution, i tried the Azure AI Document Intelligence SDK and below is what I get: Looks like the API version of the model created by AI Foundry is not being supported by the SDK. The API version created is 2024-07-31-preview. Any help on how to get around this issue?

File "C:\Users\VusirikaAi01\OneDrive - Hearst\FirstDataBank\HDA.venv\lib\site-packages\azure\ai\documentintelligence_operations_operations.py", line 819, in _analyze_document_initial

raise HttpResponseError(response=response, model=error)

azure.core.exceptions.HttpResponseError: (InvalidArgument) Invalid argument.

Code: InvalidArgument

Message: Invalid argument.

Inner error: {

"code": "InvalidParameter", "message": "The parameter CustomGenerative is invalid: Invalid build mode for this api-version."

}

PS C:\Users\VusirikaAi01\OneDrive - Hearst\FirstDataBank\HDA>
Vusirikala, Aishwarya 0 Reputation points

2025-02-10T06:35:49.5133333+00:00

Following up further:

Per Step #3, I went ahead and used the Azure AI Document Intelligence SDK and below is what i got: Looks like the API version of my model (2024-07-31-preview) is not being accepted by the SDK? How to get around with it?

File "C:\Users\VusirikaAi01\OneDrive - Hearst\FirstDataBank\HDA.venv\lib\site-packages\azure\ai\documentintelligence_operations_operations.py", line 819, in _analyze_document_initial

raise HttpResponseError(response=response, model=error)

azure.core.exceptions.HttpResponseError: (InvalidArgument) Invalid argument.

Code: InvalidArgument

Message: Invalid argument.

Inner error: {

"code": "InvalidParameter", "message": "The parameter CustomGenerative is invalid: Invalid build mode for this api-version."

}

PS C:\Users\VusirikaAi01\OneDrive - Hearst\FirstDataBank\HDA>
Vikram Singh 2,575 Reputation points Microsoft Employee

2025-02-12T11:59:46.99+00:00

Hello Vusirikala, Aishwarya,

Greetings.

Just following up to check if my suggestion helped. Please do not forget to "Accept the answer” and “up-vote” wherever the information provided helps you, this can be beneficial to other community members.

Thank you
Vikram Singh 2,575 Reputation points Microsoft Employee

2025-02-17T04:29:13.84+00:00

Hi Vusirikala, Aishwarya,

Greetings.

If the response helped, please do click Accept Answer and Yes for was this answer helpful.

Doing so would help other community members with similar issue identify the solution. I highly appreciate your contribution to the community.

Thank You.
Vikram Singh 2,575 Reputation points Microsoft Employee

2025-02-19T04:38:58.16+00:00

Hi Vusirikala, Aishwarya,

Greetings.

We haven’t heard from you on the last response and was just checking back to see if you got a chance to try above suggestions.

If the response helped, please don't forget to close up the thread here by upvoting and accept it as an answer.

Thank You.
Simon Ostermann 5 Reputation points

2025-04-16T14:53:20.74+00:00

I'm experiencing the exact same issue. The model created in AI Studio always defaults to API version 2024-07-31-preview, and I haven't found a way to change that. I'm using the Azure.AI.DocumentIntelligence .NET package, but I still encounter the same error.

Azure.RequestFailedException: Invalid argument. Status: 400 (Bad Request) ErrorCode: InvalidArgument Content: {"error":{"code":"InvalidArgument","message":"Invalid argument.","innererror":{"code":"InvalidParameter","message":"The parameter CustomGenerative is invalid: Invalid build mode for this api-version."}}}

Additionally, I can't change the API version of the SDK through DocumentIntelligenceClientOptions, since the enum only provides a single option, which is ServiceVersion.V2024_11_30.

Answer 1

Vikram Singh 2,575 Microsoft Employee

Hello Simon Ostermann,

Thanks for posting your question in Microsoft Q&A. I apologize for any inconvenience caused.

Can you please use "generative" instead of "CustomGenerative" as a parameter in your API request? This should resolve the issue you are facing.

User's image

For more info, please refer the doc.

Please do not forget to "Accept the answer" if the information provided helps you, this can be beneficial to other community members.

Thanks

Simon Ostermann 5

Hello,

Thanks for the reply!

I'm using the .NET SDK and not sure how to specify the type of the model in here:

var credential = new AzureKeyCredential(config.Value.AzureDocumentIntelligence.ApiKey);
var client = new DocumentIntelligenceClient(new Uri(config.Value.AzureDocumentIntelligence.Endpoint), credential);
var binaryData = await BinaryData.FromStreamAsync(stream);
var operation = await client.AnalyzeDocumentAsync(WaitUntil.Completed, config.Value.AzureDocumentIntelligence.Model, binaryData);
var result = operation.Value;
var document = result.Documents[0];

Vikram Singh 2,575 Reputation points Microsoft Employee

2025-04-22T04:34:41.84+00:00
Hello Simon Ostermann,

Thanks for getting back to me.

For the .NET SDK, please follow the links below and implement the modifications in your code based on the instructions provided:

DocumentBuildMode Struct (Azure.AI.DocumentIntelligence) - Azure for .NET Developers | Microsoft Learn

https://github.com/Azure/azure-sdk-for-net/blob/a491c5696ca9a18c790ac42a78db84c72af1c1cc/sdk/documentintelligence/Azure.AI.DocumentIntelligence/src/Generated/DocumentBuildMode.cs

https://github.com/Azure/azure-sdk-for-net/blob/a491c5696ca9a18c790ac42a78db84c72af1c1cc/sdk/documentintelligence/Azure.AI.DocumentIntelligence/src/Generated/DocumentBuildMode.cs

Additionally, it would be helpful if you could share the documentation that you are following or referring for the implementation.

Thanks
Simon Ostermann 5 Reputation points

2025-04-22T09:42:57.1366667+00:00

I'm a bit unclear on how the links you provided relate to what I'm trying to achieve. From what I can tell, the referenced DocumentBuildMode struct and associated files seem to be tied to a different part of the SDK than what I’m working with.

The core issue appears to be a version mismatch: the SDK version I’m using is fixed and does not support the model version 2024-07-31-preview, which is what my custom model on https://ai.azure.com is based on. Additionally, I haven't been able to find any documentation on how to use the new document field extraction capabilities with this model/version - it seems this feature is either undocumented or not yet supported in the current .NET SDK.

Just to clarify: in the code I provided earlier, I did not specify any version, build mode, or similar parameters. My goal is simply to analyze a document using an already created custom model. If possible, could you help me by modifying the existing code I shared to work with that setup?
Vikram Singh 2,575 Reputation points Microsoft Employee

2025-04-22T11:16:00.8966667+00:00

Hello Simon Ostermann,

Thank you for bringing this issue to my attention. I understand that you're experiencing difficulties with the Azure.AI.DocumentIntelligence .NET package and the API version. I will look into this matter and get back to you with a solution as soon as possible.

In the meantime, could you please provide more details about your environment and the steps you've taken to resolve the issue? This will help me better understand the problem and find an appropriate solution.

Thank you for your patience. I'll get back to you shortly.
Simon Ostermann 5 Reputation points

2025-04-22T11:38:51.6466667+00:00

I use the Document Intelligence SDK for .NET version 1.0.0 to work with models I create on https://documentintelligence.ai.azure.com/, where I can create custom neural models. However, I cannot create custom generative neural models there, so I create those on https://ai.azure.com/ using the preview document field extraction feature.

The issue arises when I try to use the SDK with models created on the second site (custom generative neural models). While the SDK works fine with models created on the first site, I get the error message mentioned above when I try to use it with generative models.
Vikram Singh 2,575 Reputation points Microsoft Employee

2025-04-23T05:45:52.29+00:00
Hello Simon Ostermann,

Thanks for sharing the info, can you try out the below step to resolve the issue:

Remove the existing package and install the prerelease version using the following command: dotnet add package Azure.AI.DocumentIntelligence --prerelease

For more detailed information, please refer to:

Azure Document Intelligence client library for .NET - version 1.0.0-beta.3

NuGet Package - Azure.AI.DocumentIntelligence

NuGet Package - Azure.AI.DocumentIntelligence 1.0.0-beta.3

Do let me know if you are still facing the issue, will investigate it further.

Thanks
Simon Ostermann 5 Reputation points

2025-04-23T12:39:04.3066667+00:00

This will install Azure.AI.DocumentIntelligence version 1.0.0, as it is the latest available release. Versions beta.1-3 are older, which you can verify here: NuGet Versions.

I am already using 1.0.0.
Vikram Singh 2,575 Reputation points Microsoft Employee

2025-04-24T05:21:40.0266667+00:00

Hello Simon Ostermann,

Could you please confirm if you are removing the existing packages and installing the prerelease version using the following command?

Thanks

Vusirikala, Aishwarya 0

Hi Vikram,

Below is my python code that I have used to integrate with Azure AI foundry Document field extraction.

        # Use the API version that matches the custom model

        api_version = "2024-07-31-preview"

        

        url = f"{self.ENDPOINT}documentintelligence/documentModels/{model_id}:analyze?api-version={api_version}&stringIndexType=utf16CodeUnit"

        

        headers = {

            "Ocp-Apim-Subscription-Key": self.API_KEY,

            "Content-Type": "application/pdf",

            "x-ms-client-request-id": str(uuid.uuid4()),

            "x-ms-useragent": "azure-ai-documentintelligence/4.0.0",

            "Cache-Control": "no-cache",

            "Accept": "application/json"

        }

Vusirikala, Aishwarya 0 Reputation points

2025-04-24T05:33:10.4566667+00:00

the issue was not with the API version, but it was the "headers" that helped me resolve my issue.

Share via

Document Text extraction model

1 answer

Your answer