Llama
Meta's open-weight language model family, available for research and commercial use, designed for efficiency and quality.
Think of it like
The open-source champion; changed the game by releasing weights under a permissive license.
Example
Llama 2 (13B and 70B versions) outperforms many closed models while being free to run locally.
How it actually works
Llama 1 (2023) leaked; Llama 2 (2023) was officially released with a commercial license (some restrictions for very large companies). Training details published openly. Spawned massive fine-tuning ecosystem.
For product teams
Open weights = huge adoption, ecosystem momentum, but harder to monetize; Meta benefits from ecosystem goodwill.
For engineers
Llama runs locally on consumer GPUs (7B fits on 8GB VRAM); foundation for Ollama, LM Studio, etc.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome