@aisavvy: Meta released Llama-3, a powerful open source AI Model that outperforms on the benchmarks models like Gemini or Claude. Llama-3 is available in 2 versions, 8B and 70B parameters. It has been trained on a dataset of 15T tokens which is 7x more than Llama-2. The context window is 8K tokens which can be extended, and the knowledge cutoff is December 2023. You can download the model or test it online on Meta AI. #ai #llama3 #metaai #aichatbot #opensource #aimodel