Meta Releases Two New AI Models as Industry Debates Evidence Standards for AI Claims
Semafor—Meta Platforms unveiled two new artificial intelligence models, continuing the company's rapid release cadence as it competes against rivals in the generative AI market. The releases come as broader debates over AI system benchmarking and the reliability of model performance claims intensify in research and policy circles. Meta has been investing heavily in open-source AI through its Llama model family while also developing proprietary systems for internal use. Critics are urging more rigorous empirical standards before AI capabilities claims are treated as settled.