Meta AI Lab Delivers First Models Amid Strategic Overhaul

Meta Platforms (META-US) has delivered its first batch of advanced AI models from a newly formed AI laboratory, marking an early milestone in the company’s restructured artificial intelligence strategy. The announcement was made by Meta’s Chief Technology Officer Andrew Bosworth during the World Economic Forum in Davos on January 21, 2026. The team, part of Meta’s Super Intelligence Lab, has been active for less than six months and has produced models with strong initial performance, though no specific model names were disclosed. The lab is reportedly developing multiple models, including the text-based “Avocado” and image/video generation project “Mango,” both expected to launch in Q1 2026. Despite these early results, Bosworth emphasized that the models remain in development and require extensive engineering before becoming usable products. This progress comes after Meta reshuffled its AI leadership and hired top-tier talent amid rising competition from Alphabet (GOOGL-US) and other tech giants. While earlier Llama 4 releases drew criticism, Meta now says its 2025 investments are beginning to yield measurable returns.