Llama
Llama is Meta's family of open-weight large language models, available for download, self-hosting, and via Meta's hosted Llama API for chat completions, text generation, and embeddings. The ecosystem also includes the open-source llama-models, llama-stack, and PurpleLlama safety projects on GitHub.
APIs
Llama API
Meta's hosted REST API providing access to Llama large language models for chat completions, text generation, and embeddings. Authentication is via API key issued through llama....
Llama Models (open weights)
Open-weight Llama model family available for direct download and self-hosted inference, distributed under Meta's Llama Community License.
Llama Stack
Open-source framework standardizing the building blocks for Llama-based generative AI applications, including inference, safety, agents, and evaluation, with REST and SDK access...
PurpleLlama
Open-source project providing tools and evaluations for assessing and improving the safety and security of generative AI models, including Llama Guard and CyberSecEval.