basics
Not every AI lab keeps its models locked away — some publish the underlying weights for anyone to download and run themselves.
Llama, the flagship of open-weight models that anyone can download and run on their own servers, was released by Meta, the parent company of Facebook and Instagram. Releasing the underlying weights publicly, rather than keeping them locked behind a paid API, let researchers and companies everywhere build on and customize the model directly, sparking a whole ecosystem of open alternatives.
Apple and Samsung are both major technology companies, but neither released Llama or built the foundational open-weight model family that ignited this movement; that credit belongs specifically to Meta.
Llama's open release helped inspire other labs to publish their own open-weight models, including Mistral, DeepSeek, and Qwen, creating real competition with closed, proprietary systems and giving developers far more choice in how they build AI applications.
basics
What do we call models that handle multiple forms of input and output?What are these 'reasoning models' doing during that pause?What's this approach called?What is it?What's one of the industry's main responses?What is this empirical rule — the justification for billion-dollar bets — called?What's the acronym for the goal beyond task-specific AI — machine intelligence as general as a human's?What are these called?What's the umbrella term?What are these pieces called?When people describe an AI model as having '70 billion parameters,' what are they counting?What is this phenomenon called?Quration — Quration AIQ