Meta has reportedly developed an artificial intelligence (AI)-powered image and video model under its superintelligence lab led by Scale AI co-founder, Alexandr Wang.
According to The Wall Street Journal, Meta's recent efforts are a part of its broader plan to build a new generation of models that specialise in image and video related tasks, along with the cutting-edge text processing, as the company seems to rival more aggressively with its competitors.
One of the models, internally codenamed “Mango,” is particularly designed to specialise in video and image comprehension.
The model aims to explain the physical world more effectively via visual data analysis and is likely to feature the advanced text-to-video generation capabilities.
Some experts believe these features have enough potential to outshine several existing offerings in the market.
Moreover, Meta is developing a new text-based large language model called “Avocado,” which is likely to deliver enhanced coding, reasoning and problem-solving capabilities.
Both Avocado and Mango are likely to be released in early 2026. Wang further confirmed that the company is currently exploring so-called “world models” that are designed to comprehend and reason about real-world environments by processing visual inputs.
The push comes as Meta has restructured its AI division after increased competition from Google, OpenAI, and Anthropic.
With no standout AI product yet, the success of these forthcoming models is seen as critical to Meta’s future AI strategy.
