Open source AI models are on their way to surpassing proprietary models, says Yann LeCun, Meta's chief AI scientist and Turing Award winner. The scatterplot shows that many open source models are already rapidly approaching or even surpassing the performance of proprietary models, such as GPT-3, Chinchilla 70B (Google), PaLM (Google), GPT-4 (OpenAI) and Llama65B (Meta). Among them, open-source models such as Llama and Yi have performed well and challenged proprietary models in terms of benchmarking, functional support and business models.
