Hello David Captur,
Welcome to the Microsoft Q&A and thank you for posting your questions here.
I understand you're having issues when attempting to use the web_search_preview tool with the Azure OpenAI Responses API.
The fact that the web_search_preview
tool is not supported in the 2025-03-01-preview
version of the Responses API. This means that any request including this tool will consistently fail, regardless of model or input, because the API simply doesn’t recognize or process that tool type at this time.
Therefore, your expectation that the Responses API can handle real-time web search through this tool is what’s causing the failure. The error message may appear to suggest a transient server-side issue, but in reality, it’s a structural incompatibility. This is not a temporary outage or misconfiguration it’s a limitation of the current API capabilities. The documentation confirms that tool support is limited in this endpoint, and web_search_preview
is not among the supported features.
To move forward reliably, the best practice is to remove the tools
section entirely from your request payload. For example:
{
"model": "gpt-4.1-mini",
"input": "Tell me today's date and a cool fact."
}
If your use case requires real-time search capabilities, the most dependable approach is to either:
- Use the Azure OpenAI Assistants API (when available in your region and subscription), which is designed to support tools like web search natively, or
- Integrate with external services such as the Bing Search API or Azure Cognitive Search, and then pass the results into your prompt manually.
To ensure you’re always working with the latest capabilities and limitations, I recommend regularly checking the official Azure OpenAI documentation: https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/responses. You can avoid these types of errors and ensure a more stable and predictable integration by by aligning your implementation with the supported features of the API.
I hope this is helpful! Do not hesitate to let me know if you have any other questions or clarifications.
Please don't forget to close up the thread here by upvoting and accept it as an answer if it is helpful.