Meta is developing a new image and video model for a 2026 release, report says

Meta is intensifying its focus on artificial intelligence as it embarks on the development of innovative models within its superintelligence lab, spearheaded by Alexandr Wang, co-founder of Scale AI. According to a recent report by The Wall Street Journal, the tech giant is working on an image and video processing model codenamed "Mango," alongside a text-based model referred to internally as "Avocado." The anticipated release of these groundbreaking models is set for the first half of 2026. During an internal Q&A session at Meta, both Wang and Chief Product Officer Chris Cox outlined the company’s strategic roadmap for these new technologies. Wang emphasized that the text-based model aims to enhance coding capabilities while also delving into advanced world models that can comprehend visual information and exhibit reasoning, planning, and action without exhaustive training on every scenario. Despite these ambitious plans, Meta has recently found itself trailing competitors like OpenAI, Anthropic, and Google in the race for AI dominance. This year has seen notable restructuring within Meta's AI division, which included leadership shifts and the recruitment of researchers from prominent firms. However, it's worth noting that several of these newly hired experts at Meta SuperIntelligence Labs (MSL) have since departed the company. Additionally, last month, Yann Lecun, the company’s chief AI scientist, announced his exit to pursue his own startup. As of now, Meta has yet to launch a flagship AI product. The performance of its AI assistant is primarily supported by the vast user base of its social networking platforms, with the assistant integrated into the search functionalities of its applications. The stakes are high for the initial outputs from MSL, as they are expected to play a crucial role in shaping Meta's future in the AI landscape.