AI research publications and industry analysis reports

Meta to Release Open-Source Llama 3 Model with 400B Parameters on July 23, Outperforming GPT-4o

Jul 22, 2024, 05:28 PM

Meta Platforms Inc. is set to release its highly anticipated Llama 3 model on Tuesday, July 23. The Llama 3 405B model, which has 400 billion parameters, is expected to be the largest open-source AI model to date. Initial benchmarks suggest that Llama 3.1 405B outperforms OpenAI's GPT-4o and Anthropic's Claude 3.5 Sonnet in various tests, although it falls short in certain areas like human evaluation and social sciences. The release comes after Meta invested $30 billion in GPUs for training the model. This launch marks a significant milestone as Llama 3 could become the first open-source model with open weights to rival proprietary models, potentially changing the landscape of AI development. Meta has already released a smaller version of Llama 3.