Meta Develops New AI Models Under Alexandr Wang

Meta Platforms is developing a new suite of software-based artificial intelligence models through its Superintelligence Labs to regain a competitive edge after Llama 4 underperformed expectations. Led by Chief AI Officer Alexandr Wang, the company is pursuing a partial open-source strategy. This approach involves releasing some model versions to developers while keeping specific components proprietary to facilitate safety evaluations and protect competitive advantages.

To support the push toward artificial general intelligence, the company invested $14.3 billion for a 49% stake in Scale AI. Meta is further bolstering its technical infrastructure by purchasing NVIDIA H100 chips to provide the computing power required to train larger, more powerful models.

Unlike competitors OpenAI and Anthropic, which prioritize enterprise and government solutions, Meta is focusing on consumer-centric AI integrated directly into WhatsApp, Facebook, and Instagram. The company aims to differentiate itself by offering open-source options rather than relying exclusively on subscription-based models, provided the new releases meet established performance benchmarks.