Meta AI Models 2026: New Image and Video Tools in Development
Meta is developing new artificial intelligence systems for a 2026 release. Reports say the company is working on advanced image and video tools. These efforts come from Meta’s superintelligence lab.
The lab is led by Scale AI co-founder Alexandr Wang. According to reports, the image and video model uses the codename Mango. In addition, Meta is building a text-based model called Avocado.
Executives shared the roadmap during an internal company session.
They plan to launch the new Meta AI models in early 2026.
As a result, expectations inside the company are high.
Why Meta’s AI strategy matters
Meta wants its text model to perform better at coding tasks. At the same time, the company is exploring world models. These systems aim to understand visuals and reason independently.
However, Meta has struggled to keep pace with rivals. Companies like OpenAI, Google, and Anthropic currently lead the AI race. Therefore, Meta faces growing pressure to deliver competitive tools.
The company’s AI division has seen major changes this year. Leadership reshuffles and new hires marked the transition. Still, several researchers have already exited the superintelligence lab.
Last month, chief AI scientist Yann LeCun announced his departure.
He plans to launch his own startup. This move added more uncertainty to Meta’s AI future.
Meta also lacks a standout AI product so far. Its assistant benefits mainly from placement inside popular apps. As a result, usage relies on Meta’s massive social audience.
The first releases from the lab carry heavy expectations. Success could restore Meta’s reputation in AI. Failure, however, may widen the gap with competitors.