GPT-4o Assistant - Invalid JSON responses breaks function calling

Issue as described in topic. Switched assistants to GPT-4o, no longer receiving valid JSON when function calls are required. This breaks my assistant flow completely.

The only change made from the last stable version was the model (gpt-4-turbo to gpt-4o)

1 Like

4o does not have a response_format parameter. https://platform.openai.com/docs/api-reference/chat/create#chat-create-response_format

You have to instruct it to produce the JSON given a schema and cross your fingers.