TapTechNews on July 13th, The Information, a tech media, released a blog post this Friday, citing an internal employee's爆料 that Meta plans to release the open-source AI model Llama3-405B on July 23rd. In addition to the existing two versions with 8 billion and 70 billion parameters, a 405 billion parameter version will be launched, claiming to be the most powerful open-source large language model.
When Meta released the Llama3 AI model in April this year, Zuckerberg revealed in an interview that it was training a 405 billion-pixel dense model, but did not disclose much information at that time.
It is reported that Llama3-405B is a multimodal open-source AI model that can understand various media contents such as images and text.
TapTechNews queried public information such as social media and found that there are already developers using this AI model.