Issue with Azure Mistral OCR Deployment – Always Returning null

Question

Issue with Azure Mistral OCR Deployment – Always Returning null

Francisco dos Santos 25

I’ve successfully deployed the Mistral OCR model via Azure AI Studio:

Deployment Name: mistral-ocr-2503-admjg

Model: mistral-ocr-2503

Endpoint: https://mistral-ocr-2503-admjg.eastus2.models.ai.azure.com/v1/ocr

Status: Succeeded

I’m sending valid PDF documents encoded in Base64 to the /v1/ocr endpoint using the image_url field, as shown in the documentation. The request is accepted with a 200 OK response, but the response body is always null, even for clean PDFs containing machine-readable text.

I’ve tested the same documents with the public Mistral API and received valid OCR results, so I suspect the issue is with the Azure-hosted deployment.

Could you please verify whether the OCR model is fully operational in my deployment? Or advise if additional configuration is needed to enable document processing?

Thanks in advance for your help.

SriLakshmi C 6,010 Reputation points Microsoft External Staff Moderator

2025-04-23T17:34:00.4666667+00:00

Hello @Francisco dos Santos,

I understand that you're experiencing an issue with the Mistral OCR model deployment in Azure, where the response is consistently null despite receiving a 200 OK status.

Here are a few suggestions to troubleshoot the problem:

Ensure that your PDF documents are properly encoded in Base64 and conform to the expected input format. Double-check that you're using the correct fields (e.g., image_url or file input) as per the model's API documentation.

The Mistral OCR model supports PDFs with up to 1,000 pages and 50MB max file size. Ensure your documents fall within these constraints.

Your Azure-hosted deployment may require specific settings that differ from the public API. Check the model deployment configuration in Azure for any custom settings related to file handling or input types.Use the Azure portal or CLI (az ml online-deployment get-logs) to review logs for errors or warnings. These can provide insight into processing issues like malformed input or unsupported file types.

Ensure your Azure subscription has sufficient quotas to handle OCR requests. Also, verify that your deployment is healthy redeploying the model may help resolve transient issues.

Kindly refer this Featured models of Azure AI model catalog.

I Hope this helps. Do let me know if you have any further queries.

Thank you!
SriLakshmi C 6,010 Reputation points Microsoft External Staff Moderator

2025-04-24T12:34:58.6733333+00:00

Hi @Francisco dos Santos,

Following up to see if the above suggestion was helpful. And, if you have any further query do let us know.

Thank you!
SriLakshmi C 6,010 Reputation points Microsoft External Staff Moderator

2025-04-25T09:31:38.5766667+00:00

Hi @Francisco dos Santos,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.

Thank you!

Samuel Weber 0

Im having the same problem. I tried sending a jpg of a pdf, and got same response, 200, Null.

Heres the code I’m using to test:

import base64
import requests
import os
def encode_image(image_path):
    """Encode the image to base64."""
    try:
        with open(image_path, "rb") as image_file:
            return base64.b64encode(image_file.read()).decode('utf-8')
    except FileNotFoundError:
        print(f"Error: The file {image_path} was not found.")
        return None
    except Exception as e:
        print(f"Error: {e}")
        return None
# Path to your image
image_path = "path_to_your_image.jpg"
# Getting the base64 string
base64_image = encode_image(image_path)
# Azure API endpoint and key
azure_endpoint = "https://mistral-ocr-2503-MyDeploymentName.eastus2.models.ai.azure.com/v1/ocr"
azure_api_key = os.environ["AZURE_API_KEY"]
# Prepare headers and payload
headers = {
    "Content-Type": "application/json",
    "Authorization": azure_api_key
}
payload = {
    "model": "mistral-ocr-2503",
    "model": "mistral-ocr-2503-deployment",  # Note: I change this to my deployment model name
    "document": {
        "type": "image",
        "image_url": f"data:image/jpeg;base64,{base64_image}"
    }
}
# Make the API request
ocr_response = requests.post(
    azure_endpoint,
    headers=headers,
    json=payload
)
# Parse the response
if ocr_response.status_code == 200:
    result = ocr_response.json()
    # Process the result as needed
else:
    print(f"Error: {ocr_response.status_code}")
    print(ocr_response.text)

Francisco dos Santos 25 Reputation points

2025-04-25T15:13:54.1433333+00:00
Hi SriLakshmi,

Thank you for the suggestions.

I’ve followed all the recommended steps to troubleshoot the issue, and I’d like to confirm the following:

The PDF files are properly encoded in Base64 using the correct data:application/pdf;base64,... format.

I’m using the correct field name: image_url, along with "type": "document" (also tested "image").

The file is under 50MB and contains just 1 page of simple machine-readable text.

The model and endpoint names in the payload match exactly what’s configured in the Azure portal.

I’ve tested both with my own PDFs and with a sample document generated specifically for OCR testing.

However, the response continues to return null with a 200 OK status — no errors, no logs, no output.
SriLakshmi C 6,010 Reputation points Microsoft External Staff Moderator

2025-04-28T14:38:56.7766667+00:00

Hi @Francisco dos Santos,

Thank you for confirming the troubleshooting steps you've already taken and for providing the detailed information.

It does appear that the behavior you're encountering is not expected. Please try below steps:

Although the documentation suggests using image_url, some OCR models deployed via Azure AI Studio may expect file inputs differently when working with Base64 content. Instead of image_url, could you try sending the payload using the file field directly with the raw Base64-encoded PDF? Some models treat image_url strictly as a URL to an image hosted externally rather than Base64 content.

If supported in the model’s endpoint, sending the document as a direct file upload (using the file field) may produce a valid output instead of using image_url. This distinction sometimes causes silent failures without error messages.

Since this is a Mistral OCR model integrated through Azure AI Foundry or Azure AI Studio, model-specific guidance might differ slightly from standard public Mistral API behavior. I recommend reviewing any model-specific guidelines published in the Azure AI Model Catalog for Mistral OCR (if available).

To rule out any deployment-specific issue, I can suggest either redeploying the model or creating a new deployment instance and testing the same document. Sometimes initial deployments can encounter silent configuration issues.

Please let me know if you have further queries.

Thank you!

Filippo Buoncompagni 0

try this

class MistralOCR:
    def __init__(self): 
        self.api_key = ""
        self.endpoint = ""
        self.model = "mistral-ocr-2503"

    def encode_pdf_as_base64(self, pdf_stream: BytesIO) -> str:
        """Encode PDF content as base64 string"""
        try:
            pdf_stream.seek(0)
            pdf_bytes = pdf_stream.read()
            base64_encoded = base64.b64encode(pdf_bytes).decode('utf-8')
            return base64_encoded
        except Exception as e:
            print(f"Error encoding PDF as base64: {e}")
            raise

    def _create_payload(self, pdf_base64: str) -> dict[str, Any]:
        return {
            "model": self.model,
            "document": {
                "document_url": f"data:application/pdf;base64,{pdf_base64}"

            },
            "include_image_base64": True
        }

    def generate_response(self, pdf_base64: str) -> Any:
        headers = {
            "Content-Type": "application/json",
            "Authorization": self.api_key,
        }

        payload = self._create_payload(pdf_base64)
        response = requests.post(self.endpoint, headers=headers, data=json.dumps(payload))
        
        response.raise_for_status()
        return response.json()

You have to put document_url and not image_url. The documentation is wrong.

SriLakshmi C 6,010 Reputation points Microsoft External Staff Moderator

2025-04-29T10:44:55.6666667+00:00
Hi @Francisco dos Santos,

Could you please try this

from azure.ai.inference import InferenceClient import os endpoint = os.environ.get("AZUREAI_ENDPOINT") api_key = os.environ.get("AZUREAI_API_KEY") client = InferenceClient(endpoint=endpoint, api_key=api_key) response = client.chat_completions.create( model="mistral-ocr-2503", document={"type": "document_url", "document_url": "your_document_url"}, include_image_base64=True ) print(response)

The key factor might be the document type— you could try something straightforward like making a REST call using curl.

curl -X POST $AZUREAI_ENDPOINT/v1/models/mistral-ocr-2503/infer \ -H "Content-Type: application/json" \ -H "Authorization: Bearer $AZURE_API_KEY" \ -d '{"document": {"type": "document_url", "document_url": "your_document_url"}, "include_image_base64": true}'

Make sure the URL you're testing with isn’t protected use a publicly accessible image instead. If every response returns a 200 status with no data, and you've only tested with files stored in secure locations, try using an image from your organization's public website or a direct link to an image with text from Bing (e.g., an example Mistral image from GitHub or Bing Copilot). This helps confirm whether the issue lies with the input or the actual endpoint configuration/deployment.

Please refer this https://docs.aimlapi.com/api-references/vision-models/ocr-optical-character-recognition/mistral-ai/mistral-ocr-latest

Thank you!
SriLakshmi C 6,010 Reputation points Microsoft External Staff Moderator

2025-04-30T09:31:50.9866667+00:00

Hi @Francisco dos Santos,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.

Thank you!
FARGETON Herve (In Extenso) 0 Reputation points

2025-06-27T13:51:34.3266667+00:00

Hello I disagree with the solution proposed. It doesn't work correctly - almost half time. Using pdf full content or only one image content, there's always a reliability problem.

Accepted answer

0 additional answers

Your answer

SriLakshmi C 6,010 Reputation points Microsoft External Staff Moderator

2025-04-23T17:34:00.4666667+00:00

Hello @Francisco dos Santos,

I understand that you're experiencing an issue with the Mistral OCR model deployment in Azure, where the response is consistently null despite receiving a 200 OK status.

Here are a few suggestions to troubleshoot the problem:

Ensure that your PDF documents are properly encoded in Base64 and conform to the expected input format. Double-check that you're using the correct fields (e.g., image_url or file input) as per the model's API documentation.

The Mistral OCR model supports PDFs with up to 1,000 pages and 50MB max file size. Ensure your documents fall within these constraints.

Your Azure-hosted deployment may require specific settings that differ from the public API. Check the model deployment configuration in Azure for any custom settings related to file handling or input types.Use the Azure portal or CLI (az ml online-deployment get-logs) to review logs for errors or warnings. These can provide insight into processing issues like malformed input or unsupported file types.

Ensure your Azure subscription has sufficient quotas to handle OCR requests. Also, verify that your deployment is healthy redeploying the model may help resolve transient issues.

Kindly refer this Featured models of Azure AI model catalog.

I Hope this helps. Do let me know if you have any further queries.

Thank you!
SriLakshmi C 6,010 Reputation points Microsoft External Staff Moderator

2025-04-24T12:34:58.6733333+00:00

Hi @Francisco dos Santos,

Following up to see if the above suggestion was helpful. And, if you have any further query do let us know.

Thank you!
SriLakshmi C 6,010 Reputation points Microsoft External Staff Moderator

2025-04-25T09:31:38.5766667+00:00

Hi @Francisco dos Santos,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.

Thank you!
Samuel Weber 0 Reputation points

2025-04-25T12:31:49.2533333+00:00

Im having the same problem. I tried sending a jpg of a pdf, and got same response, 200, Null.

Heres the code I’m using to test:

import base64 import requests import os def encode_image(image_path): """Encode the image to base64.""" try: with open(image_path, "rb") as image_file: return base64.b64encode(image_file.read()).decode('utf-8') except FileNotFoundError: print(f"Error: The file {image_path} was not found.") return None except Exception as e: print(f"Error: {e}") return None # Path to your image image_path = "path_to_your_image.jpg" # Getting the base64 string base64_image = encode_image(image_path) # Azure API endpoint and key azure_endpoint = "https://mistral-ocr-2503-MyDeploymentName.eastus2.models.ai.azure.com/v1/ocr" azure_api_key = os.environ["AZURE_API_KEY"] # Prepare headers and payload headers = { "Content-Type": "application/json", "Authorization": azure_api_key } payload = { "model": "mistral-ocr-2503", "model": "mistral-ocr-2503-deployment", # Note: I change this to my deployment model name "document": { "type": "image", "image_url": f"data:image/jpeg;base64,{base64_image}" } } # Make the API request ocr_response = requests.post( azure_endpoint, headers=headers, json=payload ) # Parse the response if ocr_response.status_code == 200: result = ocr_response.json() # Process the result as needed else: print(f"Error: {ocr_response.status_code}") print(ocr_response.text)
Francisco dos Santos 25 Reputation points

2025-04-25T15:13:54.1433333+00:00

Hi SriLakshmi,

Thank you for the suggestions.

I’ve followed all the recommended steps to troubleshoot the issue, and I’d like to confirm the following:

The PDF files are properly encoded in Base64 using the correct data:application/pdf;base64,... format.

I’m using the correct field name: image_url, along with "type": "document" (also tested "image").

The file is under 50MB and contains just 1 page of simple machine-readable text.

The model and endpoint names in the payload match exactly what’s configured in the Azure portal.

I’ve tested both with my own PDFs and with a sample document generated specifically for OCR testing.

However, the response continues to return null with a 200 OK status — no errors, no logs, no output.
SriLakshmi C 6,010 Reputation points Microsoft External Staff Moderator

2025-04-28T14:38:56.7766667+00:00

Hi @Francisco dos Santos,

Thank you for confirming the troubleshooting steps you've already taken and for providing the detailed information.

It does appear that the behavior you're encountering is not expected. Please try below steps:

Although the documentation suggests using image_url, some OCR models deployed via Azure AI Studio may expect file inputs differently when working with Base64 content. Instead of image_url, could you try sending the payload using the file field directly with the raw Base64-encoded PDF? Some models treat image_url strictly as a URL to an image hosted externally rather than Base64 content.

If supported in the model’s endpoint, sending the document as a direct file upload (using the file field) may produce a valid output instead of using image_url. This distinction sometimes causes silent failures without error messages.

Since this is a Mistral OCR model integrated through Azure AI Foundry or Azure AI Studio, model-specific guidance might differ slightly from standard public Mistral API behavior. I recommend reviewing any model-specific guidelines published in the Azure AI Model Catalog for Mistral OCR (if available).

To rule out any deployment-specific issue, I can suggest either redeploying the model or creating a new deployment instance and testing the same document. Sometimes initial deployments can encounter silent configuration issues.

Please let me know if you have further queries.

Thank you!
Filippo Buoncompagni 0 Reputation points

2025-04-29T09:29:21.68+00:00

try this

class MistralOCR: def __init__(self): self.api_key = "" self.endpoint = "" self.model = "mistral-ocr-2503" def encode_pdf_as_base64(self, pdf_stream: BytesIO) -> str: """Encode PDF content as base64 string""" try: pdf_stream.seek(0) pdf_bytes = pdf_stream.read() base64_encoded = base64.b64encode(pdf_bytes).decode('utf-8') return base64_encoded except Exception as e: print(f"Error encoding PDF as base64: {e}") raise def _create_payload(self, pdf_base64: str) -> dict[str, Any]: return { "model": self.model, "document": { "document_url": f"data:application/pdf;base64,{pdf_base64}" }, "include_image_base64": True } def generate_response(self, pdf_base64: str) -> Any: headers = { "Content-Type": "application/json", "Authorization": self.api_key, } payload = self._create_payload(pdf_base64) response = requests.post(self.endpoint, headers=headers, data=json.dumps(payload)) response.raise_for_status() return response.json()

You have to put document_url and not image_url. The documentation is wrong.
SriLakshmi C 6,010 Reputation points Microsoft External Staff Moderator

2025-04-29T10:44:55.6666667+00:00

Hi @Francisco dos Santos,

Could you please try this

from azure.ai.inference import InferenceClient import os endpoint = os.environ.get("AZUREAI_ENDPOINT") api_key = os.environ.get("AZUREAI_API_KEY") client = InferenceClient(endpoint=endpoint, api_key=api_key) response = client.chat_completions.create( model="mistral-ocr-2503", document={"type": "document_url", "document_url": "your_document_url"}, include_image_base64=True ) print(response)

The key factor might be the document type— you could try something straightforward like making a REST call using curl.

curl -X POST $AZUREAI_ENDPOINT/v1/models/mistral-ocr-2503/infer \ -H "Content-Type: application/json" \ -H "Authorization: Bearer $AZURE_API_KEY" \ -d '{"document": {"type": "document_url", "document_url": "your_document_url"}, "include_image_base64": true}'

Make sure the URL you're testing with isn’t protected use a publicly accessible image instead. If every response returns a 200 status with no data, and you've only tested with files stored in secure locations, try using an image from your organization's public website or a direct link to an image with text from Bing (e.g., an example Mistral image from GitHub or Bing Copilot). This helps confirm whether the issue lies with the input or the actual endpoint configuration/deployment.

Please refer this https://docs.aimlapi.com/api-references/vision-models/ocr-optical-character-recognition/mistral-ai/mistral-ocr-latest

Thank you!
SriLakshmi C 6,010 Reputation points Microsoft External Staff Moderator

2025-04-30T09:31:50.9866667+00:00

Hi @Francisco dos Santos,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.

Thank you!
FARGETON Herve (In Extenso) 0 Reputation points

2025-06-27T13:51:34.3266667+00:00

Hello I disagree with the solution proposed. It doesn't work correctly - almost half time. Using pdf full content or only one image content, there's always a reliability problem.

Answer 1

Arthur-Olivier Fortin 140

What worked for me :

{
	"model": "mistral-ocr-2503-deployment",
 	"document": 
	{    
		"document_url": "data:image/jpeg;base64,{your_base_64_image_data}",
	    "type": "image_url"  
	},   
	"include_image_base64": true
}

for pdfs you need to change the type param for "document_url"

The documentation should be updated, it is quite misleading.

Share via

Issue with Azure Mistral OCR Deployment – Always Returning null

0 additional answers

Your answer