Not an OpenAI wrapper in fact. Using multiple models but there is a large amount of code above the models. Read the ArXiv paper and you will understand more. https://arxiv.org/abs/2403.02164
EDIT: a very interesting paper since I am specifically interested in integrating neuro-symbolic cognition with large language models. I was paid to work, generally, in the field of symbolic AI in the 1980s and early 1990s and in retrospect I feel like I wasted a lot of my time. I think it is likely that symbolic AI will make a comeback if combined with LLMs, LLMs+function calling, integration with 'explain your steps' LLM prompts, etc.