openai.sync and similar wrappers have a bug with not passing the model into generate in a smart way - also affects the auto llm wrapper
openai.sync and similar wrappers have a bug with not passing the model into generate in a smart way