Browse Models
Llama 3 is a family of transformer-based large language models developed by Meta, first released in April 2024, encompassing models ranging from 3B to 70B parameters with variants including Llama 3.1, 3.2, and 3.3. These decoder-only models feature grouped-query attention, 128K token vocabulary, extended context windows up to 128,000 tokens, and support for eight languages including English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai.