Meta Llama 3.1 405B: Meta launched its largest AI model, has passed 150 tests

Meta launched its biggest AI model Llama 3.1 405B. This is the largest artificial intelligence-based large language model by the company. Meta CEO Mark Zuckerberg has also claimed that this is an open-sourced model that will compete with the likes of GPT-4, GPT-4o, and Claude 3.5 Sonnet. Meta has also upgraded the previously launched Llama 3 8B and 70B AI.

The new version supports 1,28,000 tokens. It is operable in general knowledge, operability, mathematics, device usage, and translation into many languages. 405B works on 405 billion parameters which is the model with the highest number of parameters to date. This model supports English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai languages.

Before launching Llama 3.1 405B, 150 benchmark tests were conducted, and it passed. This AI model of Meta has scored 96.8 points in mathematics, GPT-4 has bagged 94.2 points, GPT-4o has scored 96.1 and Claude 3.5 Sonnet took 96.4 points.

They also trained an AI model in Meta of more than 15 trillion tokens using more than 16 thousand class Nvidia H100 GPUs. That was also available on clouds like AWS, Nvidia, Databricks, Groq, Dell, Azure, Google Cloud, and Snowflake.