Translated data: Meta is reportedly developing a more powerful AI large model, Llama3, which is expected to be open-source and free. It is said that Llama3 aims to achieve performance levels comparable to GPT-4, although this has not been officially confirmed by Meta. Currently, Llama2 has reached the performance level of GPT-3.5 in certain applications, but it still has a noticeable gap compared to models like GPT-4. Whether Llama3 can surpass GPT-4 technically remains uncertain.