Official benchmark reports from Meta and OpenAI

Meta Releases Llama 3.1 405B, Largest Open-Source AI Model, Outperforms GPT-4o

Jul 23, 2024, 01:14 PM

Meta has officially released Llama 3.1, a new open-source AI model with 405 billion parameters, making it the largest and most powerful open-source model to date. The model, which also includes 8B and 70B versions, supports multiple languages and boasts a 128K context length. Leaked benchmarks suggest that Llama 3.1 405B outperforms OpenAI's GPT-4o in various metrics, except for HumanEval and MMLU_social_sciences. The release marks a significant milestone in AI as it positions Meta's open-source model to rival leading proprietary models. The model was leaked on 4chan before its official release.