This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(182.40000000000003, 71%, 27%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EAO%3C/text%3E%3C/svg%3E)
I am requesting from o4-mini to find the dimensions of a technical drawing of a cnc machined part. it fails with 'Forbidden'. I get this error: Error: 403 Your resource has been temporarily blocked because we detected unusual behavior.
Hi Ali Oztan,
To help troubleshoot the issue you're facing with the "Forbidden" error, could you please clarify the following:
1.How are you making the request? Are you using a script, a tool like Postman, a browser, or some other method?
2.What is the exact endpoint or URL you’re trying to access? This will help me understand what resource is being blocked.
3.When you say you're trying to "find the dimensions of a technical drawing," could you please elaborate?
Are you trying to extract dimensions from an image, PDF, or CAD file? Or are you querying a system/database for this information?
Thanks! This will help me investigate the issue more accurately.
Hi Ali Oztan,
We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet.
Can you please share the asked details to investigate further.
Thank you!
1- we tried making the request from AWS lambda functions and our local computers.
2- this is the endpoint "https://secki-ma2mjbt6-swedencentral.cognitiveservices.azure.com/"
3- we are trying to extract the bounding box dimensions for a given technical drawing image(png file) of a cnc machined component.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(288, 77%, 37%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ESS%3C/text%3E%3C/svg%3E)
Hi @Prashanth Veeragoni , any updates on this?
1- we tried making the request from AWS lambda functions and our local computers.
2- this is the endpoint "https://secki-ma2mjbt6-swedencentral.cognitiveservices.azure.com/"
3- we are trying to extract the bounding box dimensions for a given technical drawing image(png file) of a cnc machined component.
Hi Ali Oztan,
Thanks for the detailed explanation. A 403 Forbidden error from an Azure OpenAI or Cognitive Services endpoint usually indicates one of the following:
Root Causes of 403 Forbidden from Azure Cognitive Services
Here are likely causes and resolutions based on your context:
1.Resource Temporarily Blocked Due to Unusual Behavior
Azure says: "Your resource has been temporarily blocked because we detected unusual behavior."
This happens when Azure detects automated abuse patterns (e.g., scraping-like behavior or too many concurrent requests without proper headers, or unsupported input).
Resolution:
· Wait for a cooldown period (typically a few hours to a day).
· Reduce request frequency.
· Ensure headers like User-Agent and Ocp-Apim-Subscription-Key are properly set.
2.Missing or Invalid Authentication Headers
Ensure your requests contain:
Ocp-Apim-Subscription-Key: <your-azure-subscription-key>
Content-Type: application/json (or multipart/form-data for images)
If you're using a Python SDK or REST API directly, double-check that the key and endpoint match the resource (like Computer Vision, Form Recognizer, or a custom model if you deployed one).
3.You Might Be Hitting a Wrong API or Region
You're using:
https://secki-ma2mjbt6-swedencentral.cognitiveservices.azure.com/
Make sure:
· The resource name secki-ma2mjbt6 exists and is of the correct service (e.g., Computer Vision or Custom Vision).
· The model you're calling (e.g., for bounding box detection) is available in Sweden Central region.
· If calling OpenAI services, they are usually under a different subdomain like:
https://<resource-name>.openai.azure.com/
4.You're Calling an Endpoint That Does Not Support That Input
If you're sending PNG files for bounding box extraction:
· Use Form Recognizer (for documents with structure) or Custom Vision for object detection.
· Make sure the model you're calling was trained for technical drawings/CAD/CNC parts — default models may not support this.
Hope this helps, do let me know if you have any further queries.
Thank you!