No able to send a request to Flux 2.0 PROD Foundry -

Question

No able to send a request to Flux 2.0 PROD Foundry -

Angie Becerra 0

I tried many times to send a request to Founday Flux 2.0 PROd without luck, I follow this code:
import base64

from openai import OpenAI

endpoint = "https://llm-hook-ideas.openai.azure.com/openai/v1/"

deployment_name = "FLUX.2-pro"

api_key = "<your-api-key>"

client = OpenAI(

base_url=endpoint,

api_key=api_key

)

img = client.images.generate(

model=deployment_name,

prompt="A cute baby polar bear",

n=1,

size="1024x1024",

)

image_bytes = base64.b64decode(img.data[0].b64_json)

with open("output.png", "wb") as f:

f.write(image_bytes)

but it not working, someone know that right way to make a model available as endpoint from Foundry thank you

Shree Hima Bindu Maganti 6,880 Reputation points Microsoft External Staff Moderator

2026-03-22T20:09:29.48+00:00
Hi @Angie Becerra
It sounds like you’ve got a working Flux 2.0 “pro” model deployed in AI Foundry but your Python code isn’t actually hitting the right endpoint or using the right SDK settings for Azure. Here’s a checklist and a sample snippet that should get you over the line:

Grab the exact endpoint and deployment name from the Azure portal – In your AI Foundry resource, go to “Keys & Endpoint” and copy the “Endpoint” URL. – In the “Deployments” blade, copy the Deployment name (it’s case-sensitive).

Don’t include /openai/v1 in your base URL – Your base URL should look like https://<your-resource-name>.openai.azure.com – The SDK will append the right route (e.g. /openai/images/generations).

Specify the API version and API type in your code – Azure’s image-generation endpoint needs an api-version query. – If you’re using the official OpenAI Python package (v0.27+), set these before you call the client.

Use the proper client constructor for Azure – Option A: openai-python SDK

import openai openai.api_type = "azure" openai.api_base = "https://llm-hook-ideas.openai.azure.com" openai.api_version = "2023-08-01-preview" # or the version shown in portal openai.api_key = "<your-api-key>" response = openai.Image.create( engine="FLUX-2-pro", # use exact deployment name prompt="A cute baby polar bear", n=1, size="1024x1024" ) img_data = response.data[0].b64_json

– Option B: azure.ai.openai SDK

from azure.ai.openai import OpenAIClient from azure.core.credentials import AzureKeyCredential import base64 client = OpenAIClient( endpoint="https://llm-hook-ideas.openai.azure.com", credential=AzureKeyCredential("<your-api-key>") ) result = client.get_images( deployment_id="FLUX-2-pro", prompts=["A cute baby polar bear"], size="1024x1024", n=1 ) image_bytes = base64.b64decode(result.data[0].b64_json) with open("output.png", "wb") as f: f.write(image_bytes)

Verify your deployment actually supports image generation – If you used a text‐only Flux model, .images.generate will 404 or “Operation Not Found.” – In the portal, make sure your deployment type is an image-generation model (like DALL·E style).

Hope this helps! If you still hit errors, can you share:

• The exact error message (404? auth error? etc.)

• Which Python/openai package version you’re using

• The full deployment name as shown in the portal
Praveen Bandaru 10,900 Reputation points Microsoft External Staff Moderator

2026-03-22T20:20:18.7066667+00:00
Hello Angie Becerra

It sounds like you’ve got a working Flux 2.0 “pro” model deployed in AI Foundry but your Python code isn’t actually hitting the right endpoint or using the right SDK settings for Azure. Here’s a checklist and a sample snippet that should get you over the line:

Grab the exact endpoint and deployment name from the Azure portal

In your AI Foundry resource, go to “Keys & Endpoint” and copy the “Endpoint” URL.

In the “Deployments” blade, copy the Deployment name (it’s case-sensitive).

Don’t include /openai/v1 in your base URL

Your base URL – Your base URL should look like https://<your-resource-name>.openai.azure.com

The SDK will append the right route (e.g. /openai/images/generations).

Specify the API version and API type in your code

Azure’s image-generation endpoint needs an api-version query.

If you’re using the official OpenAI Python package (v0.27+), set these before you call the client.

Use the proper client constructor for Azure

Option A: openai-python SDK

import openai openai.api_type = "azure" openai.api_base = "https://llm-hook-ideas.openai.azure.com" openai.api_version = "2023-08-01-preview" # or the version shown in portal openai.api_key = "<your-api-key>" response = openai.Image.create( engine="FLUX-2-pro", # use exact deployment name prompt="A cute baby polar bear", n=1, size="1024x1024" ) img_data = response.data[0].b64_json

Option B: azure.ai.openai SDK

from azure.ai.openai import OpenAIClient from azure.core.credentials import AzureKeyCredential import base64 client = OpenAIClient( endpoint="https://llm-hook-ideas.openai.azure.com", credential=AzureKeyCredential("<your-api-key>") ) result = client.get_images( deployment_id="FLUX-2-pro", prompts=["A cute baby polar bear"], size="1024x1024", n=1 ) image_bytes = base64.b64decode(result.data[0].b64_json) with open("output.png", "wb") as f: f.write(image_bytes)

Verify your deployment actually supports image generation

If you used a text‐only Flux model, .images.generate will 404 or “Operation Not Found.”

In the portal, make sure your deployment type is an image-generation model (like DALL·E style).

Hope this helps! If you still hit errors, can you share:

The exact error message (404? auth error? etc.)

Which Python/openai package version you’re using?

The full deployment name as shown in the portal?

Kindly check the below reference documents for more understanding:

Region support in Azure AI Foundry:

Microsoft Foundry SDKs and Endpoints

Deploy models as serverless API

I hope the above answer helps you! Please let us know if you have any further questions.

Please don't forget to "upvote" where the information provided will help you, this can be beneficial to other members of the community.

Angie Becerra 0

Hi Shree Hima Bindu Maganti, I tried both ways without luck:

My check list:

Model name: Flux.2-pro
Version, I tried (1, 2025-12-08, 2024-10-21) same result for all
SDK endpoint: https://llm-hook-ideas.services.ai.azure.com/api/projects/proj-feelim-hooks
Ensure model creates image: using the playground model is working, but outside .. I am getting the 404 error.

imagen

Functions:
        generate_hook_image:  http://localhost:7071/api/generate_hook_image
For detailed output, run func with --verbose flag.
[2026-03-23T15:46:36.409Z] Host lock lease acquired by instance ID '00000000000000000000000029616822'.
[2026-03-23T15:46:40.799Z] Executing 'Functions.generate_hook_image' (Reason='This function was programmatically called via the host APIs.', Id=8a762c7a-3298-449e-a077-328ecbe527df)
[2026-03-23T15:46:40.987Z] Intentando generar imagen en el deployment: FLUX.2-pro
[2026-03-23T15:46:42.700Z] HTTP Request: POST https://llm-hook-ideas.openai.azure.com/openai/deployments/FLUX.2-pro/images/generations?api-version=2024-10-21 "HTTP/1.1 404 Not Found"
[2026-03-23T15:46:42.707Z] Error detallado: NOT FOUND
[2026-03-23T15:46:42.762Z] Executed 'Functions.generate_hook_image' (Succeeded, Id=8a762c7a-3298-449e-a077-328ecbe527df, Duration=1999ms)
[2026-03-23T15:47:36.957Z] Executing 'Functions.generate_hook_image' (Reason='This function was programmatically called via the host APIs.', Id=02957e53-5d0f-4ff3-b1c6-eb164533ae2c)
[2026-03-23T15:47:36.997Z] Intentando generar imagen en el deployment: FLUX.2-pro
[2026-03-23T15:47:37.749Z] HTTP Request: POST https://llm-hook-ideas.openai.azure.com/openai/deployments/FLUX.2-pro/images/generations?api-version=2024-10-21 "HTTP/1.1 404 Not Found"
[2026-03-23T15:47:37.755Z] Error detallado: NOT FOUND
[2026-03-23T15:47:37.762Z] Executed 'Functions.generate_hook_image' (Succeeded, Id=02957e53-5d0f-4ff3-b1c6-eb164533ae2c, Duration=807ms)
[2026-03-23T15:47:56.791Z] Worker process started and initialized.
[2026-03-23T15:47:58.463Z] Executing 'Functions.generate_hook_image' (Reason='This function was programmatically called via the host APIs.', Id=01f6a5cf-d5d1-4377-ae86-59b30c3b37a0)
[2026-03-23T15:47:58.608Z] Llamando a FLUX en el deployment: FLUX.2-pro
[2026-03-23T15:47:59.729Z] HTTP Request: POST https://llm-hook-ideas.openai.azure.com/openai/deployments/FLUX.2-pro/images/generations?api-version=1 "HTTP/1.1 404 Resource Not Found"
[2026-03-23T15:47:59.738Z] Error detallado en la generación: Error code: 404 - {'error': {'code': '404', 'message': 'Resource not found'}}
[2026-03-23T15:47:59.741Z] Executed 'Functions.generate_hook_image' (Succeeded, Id=01f6a5cf-d5d1-4377-ae86-59b30c3b37a0, Duration=1278ms)
[2026-03-23T15:48:01.800Z] Host lock lease acquired by instance ID '00000000000000000000000029616822'.

Then, I created this new version without luck

My code:

import azure.functions as func
import json
import os
import logging
from openai import AzureOpenAI # Usamos el constructor específico para Azure
app = func.FunctionApp()
@app.route(route="generate_hook_image", auth_level=func.AuthLevel.ANONYMOUS)
def generate_hook_image(req: func.HttpRequest) -> func.HttpResponse:
    try:
        body = req.get_json()
        hook = body.get("hook", "Image for post")
        # 1. Configuración del Cliente según el Foro:
        # Endpoint: https://llm-hook-ideas.openai.azure.com (Sin /openai/v1 aquí)
        # El SDK moderno de AzureOpenAI agregará automáticamente la ruta correcta.
        client = AzureOpenAI(
            azure_endpoint=os.environ["AZURE_PROJECT_ENDPOINT"].strip("/ "),
            api_key=os.environ["AZURE_PROJECT_KEY"].strip(),
            api_version="1" # Versión estable para imágenes
        )
        deployment_name = os.environ.get("FLUX_DEPLOYMENT_NAME", "FLUX.2-pro")
        logging.info(f"Llamando a FLUX en el deployment: {deployment_name}")
        # 2. Generación de imagen
        # Importante: Para Azure, 'model' debe ser el Deployment Name exacto
        response = client.images.generate(
            model=deployment_name, 
            prompt=f"Professional LinkedIn illustration. Topic: {hook}. Corporate style, clean design.",
            n=1,
            size="1024x1024",
            response_format="b64_json"
        )
        # 3. Extraer el Base64
        image_b64 = response.data[0].b64_json
        return func.HttpResponse(
            json.dumps({
                "status": "success",
                "image_b64": image_b64
            }),
            mimetype="application/json",
            status_code=200
        )
    except Exception as e:
        logging.error(f"Error detallado en la generación: {str(e)}")
        # Si sigue dando 404, el problema es que el nombre del Deployment no es 'FLUX.2-pro'
        return func.HttpResponse(
            json.dumps({"error": str(e)}),
            status_code=500,
            mimetype="application/json"
        )

Please let me know my error, thank you

1 answer

Your answer

Answer 1

Hi Angie Becerra

Sorry for the delay in response.

Here is the working Curl and SDK command for reference (Tested before sharing).

Please replace "foundryhubname" and AZURE_API_KEY in below codes.

Note: Curl commands are there in endpoint itself. We can draft Python SDK based on Curl commands codes.

User's image

curl -X POST "https://<foundryhubname>.services.ai.azure.com/providers/blackforestlabs/v1/flux-2-pro?api-version=preview" \   -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AZURE_API_KEY" \
  -d '{
        "prompt" : "A photograph of a red fox in an autumn forest",
        "width" : 1024,
        "height" : 1024,
        "n" : 1,
        "model": "FLUX.2-pro"
    }' | jq -r '.data[0].b64_json' | base64 --decode > generated_image.png

import os
import base64
import requests

# Your Foundry resource endpoint (same as curl)
ENDPOINT = (
    "https://testdegradeclassic.services.ai.azure.com/"
    "providers/blackforestlabs/v1/flux-2-pro?api-version=preview"
)

# The curl used: Authorization: Bearer $AZURE_API_KEY
AZURE_API_KEY = "<apikey>"

payload = {
    "prompt": "A photograph of a red fox in an autumn forest",
    "width": 1024,
    "height": 1024,
    "n": 1,
    "model": "FLUX.2-pro",
}

headers = {
    "Content-Type": "application/json",
    "Authorization": f"Bearer {AZURE_API_KEY}",
}

resp = requests.post(ENDPOINT, headers=headers, json=payload, timeout=180)
resp.raise_for_status()

data = resp.json()
b64_img = data["data"][0]["b64_json"]  # same jq path: .data[0].b64_json
img_bytes = base64.b64decode(b64_img)

out_path = "generated_image.png"
with open(out_path, "wb") as f:
    f.write(img_bytes)

print(f"Saved: {out_path}")

Please let us know if it works at your side too.

Thank you.

Manas Mohanty 15,795 Reputation points Microsoft External Staff Moderator

2026-03-25T19:42:18.1033333+00:00

Hi Angie Becerra

Can you share your observation with respect to above code.

Thank you.

Share via

No able to send a request to Flux 2.0 PROD Foundry -

1 answer

Your answer