What is Llama
Meta’s family of open-weight language models. Anyone can download them and run them on hardware they control.
In plain terms
Llama is the best-known open-weight model family. Meta trains the models and publishes the trained files, so a company can download one and run it on its own computers instead of calling a paid API. Teams choose Llama when data cannot leave their network, when the per-request cost of an API is too high, or when they want to adjust the model themselves.
What it tells you about a developer
Llama on a profile points to a developer who has run a model rather than only called one: serving it, sizing the hardware, and watching cost and speed. That work sits closer to infrastructure than an API integration does.
Facts
- Type
- Service
- Released
- 2023
- Built by
- Meta
- Used in
- AI & machine learning
- Applied to
- AI agents & tooling