Questions about Llama (language model)

Short answers, pulled from the story.

When did Meta AI publish the first version of Llama?

Meta AI published the first version of Llama on the 24th of February 2023. The announcement revealed a model family ranging from 1 billion to 65 billion parameters.

What happened to the original Llama weights after they appeared online on the 3rd of March 2023?

A torrent containing the full weights of the original Llama appeared online on the 3rd of March 2023 and spread rapidly through communities on the 4chan imageboard. Meta filed takedown requests against HuggingFace on the 6th of March, labeling it unauthorized distribution.

How many parameters does the latest Llama 4 model have and when was it released?

Llama 4 arrived in April 2025 as the latest generation of the series. This model shifted to a mixture of experts architecture and introduced multimodal capabilities handling both text and image inputs across twelve languages.

Why do some organizations classify Llama as non-free software instead of open source?

The Free Software Foundation classified Llama 3.1 as nonfree software in January 2025 due to acceptable use policies that limited popular applications. Restrictions blocked usage by individuals within the European Union and entities with over 700 million daily active users.

Which institutions used Llama models for medical or space applications?

Meditron emerged from École Polytechnique Fédérale de Lausanne and Yale School of Medicine as a family fine-tuned on clinical guidelines and PubMed papers. Booz Allen Hamilton deployed Llama 3.2 aboard the International Space Station under the Space Llama project.