Llama (Meta AI model)
Japanese Llama（MetaのAIモデル）
Llama is a family of AI language models made by Meta. You can download the model weights and run them on your own computer or server, but Meta's license conditions apply, so read them first.
Llama is an AI model that reads and writes text like ChatGPT, but you can download the model itself, the weights file, and run it yourself. Think of it as a recipe you take home rather than a meal at a restaurant: you can cook it in your own kitchen, but you must keep the promises written on the recipe. Llama 3.1 (8B, 70B and 405B) came out on July 23, 2024, and Llama 4 on April 5, 2025.
Watch out Meta called Llama open source, but OSI says it is not, because the license limits who may use it and for what. That does not mean you cannot use it, but there are conditions, such as showing "Built with Llama" when you share something made with it. If you want to try it on your own computer, start with the smallest 8B model.
Source: Introducing Llama 3.1: Our most capable models to date · The Llama 4 herd: The beginning of a new era of natively multimodal AI innovation · Llama 3.1 Community License Agreement · Llama 4 Community License Agreement · Meta's LLaMa license is still not Open Source · llama3.1 (checked 2026-10-26)