Meta Platforms Inc (NASDAQ:META, XETRA:FB2A, SIX:FB) is developing a new image- and video-focused artificial intelligence model code-named Mango, alongside a text-based large language model called Avocado, according to a Wall Street Journal report.
Both models are expected to be released in the first half of 2026.
Per the report, Meta’s Chief AI Officer, Alexandr Wang, discussed the projects during an internal Q&A with Chief Product Officer Chris Cox.
Wang reportedly said Avocado will focus on improving coding capabilities, while the company is exploring so-called “world models,” which allow AI to learn about its environment through visual input.
The initiative follows a major restructuring of Meta’s AI team over the summer. Wang now leads the newly created Meta Superintelligence Labs, and CEO Mark Zuckerberg has recruited more than 20 researchers from OpenAI, expanding the team with over 50 new AI specialists.
Image generation has become a key area of competition in AI development. In September, Meta launched an AI video generator called Vibes, created in collaboration with the startup Midjourney.
OpenAI quickly responded with its own video generator, Sora.
Mango aims to compete with tools like Google’s Gemini models and OpenAI’s offerings, delivering high-quality creative outputs for image and video.
According to Wedbush analysts, Meta is preparing for a major AI push in the first half of 2026.
“We see model development as necessarily in part driving hardware demand, particularly with GhatGPT 5 and Gemini 3 seemingly raising the bar for models, with the former's release coinciding with a pickup in inference workloads,” they wrote.
“The question in our view is will Meta be able to move the needle as well with its efforts.”