Context: imagine you create a tool to get the weather. While it works well with all models, when switching to the models gpt-4o-realtime-preview the webhook is calling but the answer is not parsed well. The default vapi response:

{results: [{toolCallId: "call_12345678", result: "Temperature is 63F in Miami, FL"}]}

is not working. I tried also with openai format with no luck:

{results: [{tool_call_id: "call_12345678", content: "Temperature is 63F in Miami, FL", role: "tool", name: "retrieve_weather"}]}

Any idea when/if it can be fixed?

Heycquilboss Thank you for creating the ticket! A member of our team will reach out to you shortly.

@cquilboss I'm sorry to hear that you're experiencing issues with the webhook response parsing for the gpt-4o-realtime-preview model. It seems like there might be a discrepancy between the expected response format and what the model is able to parse. To address this, I would recommend checking the documentation for the gpt-4o-realtime-preview model to ensure that the response format aligns with the model's requirements. If the documentation does not resolve the issue, it would be best to reach out to the support team with details of the problem, including the response formats you've tried and the specific errors or issues with parsing. In the meantime, if there's anything else I can assist you with or if you have further questions, please let me know!

any idea?

@cquilboss can you share the call ID so I can take a look?

6824e749-392c-44aa-b883-d6fba35be69a