Microsoft paper accidentally "exposed", OpenAI parameters leaked! 4o only 200B, o1 is only 300B!

Microsoft University of Washington medical paper accidentally exposed the GPT-4, GPT-4o, o1 series of model parameters, shockingly, the GPT-4o mini is only 8B. some users speculate that the 4o mini is a MoE model with about 40B parameters, of which the activation parameter is 8B. the paper disclosed the parameters of each model are as follows: the GPT-4 parameters are about 1.76 trillion; GPT-4o parameters are about 200 billion; GPT-4o mini parameters are about 8 billion; o1-preview parameters are about 300 billion; o1-mini parameters are about 100 billion; and Claude 3.5 Sonnet parameters are about 175 billion. (NIC)