Meta AI / FAIR

Meta's AI research division — home of Yann LeCun, creator of the Llama open-source model family and PyTorch.

Meta AI (formerly Facebook AI Research, or FAIR) is Meta's AI research arm, led at the research level by Chief AI Scientist Yann LeCun (a Turing Award winner for deep-learning fundamentals). Meta also has a separate Generative AI org focused on shipping products like Meta AI assistant, AI Studio, and the Llama series. It matters most for one reason: Llama. By open-weighting Llama 1 (2023), Llama 2, Llama 3, and Llama 4, Meta single-handedly created the modern open-source LLM ecosystem. Almost every locally-runnable model — Mistral, Qwen, DeepSeek, fine-tunes from a thousand startups — traces its lineage or technique back to Llama. Mark Zuckerberg's strategy: open-source models commoditize the model layer so platforms like Meta's are the value capture point. Key contributions: Llama family (open-weight LLMs), PyTorch (now the dominant ML framework), Segment Anything (image segmentation), Code Llama, MusicGen, Voicebox, and a long history of foundational research papers. FAIR has been one of the most academically prolific AI labs. The trade-off: Meta's frontier-product execution has been bumpier — Galactica was withdrawn in 2022 after factual errors, the chatbot rollout has been mixed, and Llama 4 was met with skepticism due to suspected benchmark issues. But the open-source bet has been unambiguously historic. Related: Llama family, Yann LeCun, PyTorch, open-source.