A recent Microsoft research paper has inadvertently disclosed the parameter sizes of several AI models, including OpenAI's. The findings have sparked discussions about model architecture and performance in the industry, particularly relating to medical AI evaluation.