There are subtile differences like now it has the role "tool_call" and before it was type "tool-call", but these things made the api calls crash. Does anyone has any clue of why this might've happened? If not, only possible explanation is a change in vapi format for messages, but seems odd and it wouldn't be nice to know that these changes might come up like this. This happenend with no warnings and overnight.

@Jean Reinhold I'm sorry to hear that you've encountered issues with your API calls. It sounds like there may have been a change in the API specification that wasn't properly communicated. Subtle differences in parameter names, such as "tool_call" versus "tool-call," can indeed cause API calls to fail if the API is expecting a specific format. To address this, I would recommend reviewing the latest API documentation for any updates or changes that might have been made to the parameter names or formats. If the documentation does not clarify the changes, or if it hasn't been updated to reflect these differences, please reach out to support for more detailed information on this matter. In the meantime, if you have any specific error messages or codes, I can help you interpret them and provide guidance on how to proceed. Please provide any additional details you have, and I'll do my best to assist you.