Llama is Meta's family of open-source large language models — freely downloadable, modifiable, and deployable on your own hardware or cloud infrastructure. Llama 3.1 and 3.3 compete with frontier closed models on many benchmarks while remaining permissively licensed for commercial use. The open-source nature has spawned an enormous ecosystem: Ollama makes running Llama locally trivial, Groq offers the fastest Llama inference, and thousands of fine-tuned variants serve specific domains. For applications where data cannot leave your environment, self-hosted Llama provides frontier-level capability without API calls to external services. Llama models are the foundation of a large fraction of the AI application ecosystem. Understanding what they can do — and running them yourself — is increasingly baseline knowledge for AI practitioners.