It???s all hands on deck at Meta, as the company develops new AI models under its superintelligence lab led by Scale AI co-founder, Alexandr Wang. The company is now working on an image and video model codenamed ???Mango??? along with a new text-based model internally known as ???Avocado,??? The Wall Street Journal reported.
It’s all hands on deck at Meta as the company accelerates development of new artificial intelligence models under its recently formed Superintelligence Lab, led by Scale AI co-founder Alexandr Wang.
According to a report by The Wall Street Journal, Meta is currently working on:
These models are expected to be released in the first half of 2026, based on details shared during an internal Q&A session at Meta last week.
The roadmap was unveiled during an internal discussion featuring Alexandr Wang and Meta’s Chief Product Officer Chris Cox. The presentation outlined Meta’s renewed push to regain ground in the highly competitive AI landscape.
Wang explained that Meta’s goals include:
This represents a strategic shift toward more general-purpose and agentic AI systems.
Meta has increasingly lagged behind rivals such as OpenAI, Anthropic, and Google, all of which have made rapid progress in foundation models, multimodal AI, and enterprise adoption.
In response, Meta restructured its AI division earlier this year. These changes included:
However, the effort has not been without setbacks. Several researchers who joined Meta Superintelligence Labs (MSL) have already exited the company, raising questions about internal stability.
Adding to the turbulence, Meta’s long-time Chief AI Scientist Yann LeCun announced last month that he is leaving the company to launch his own startup.
LeCun’s departure marks a significant moment for Meta, given his influential role in shaping the company’s AI research philosophy over the past decade.
Despite heavy investment, Meta has yet to produce a standalone AI product that clearly leads the market.
The company’s Meta AI assistant benefits largely from distribution rather than demandΓÇöit is embedded directly into the search bars of Meta’s apps, giving it exposure to billions of users across Facebook, Instagram, and WhatsApp.
While this boosts usage numbers, it does not necessarily reflect competitive strength or user preference compared to rivals’ AI offerings.
As a result, the first models emerging from Meta Superintelligence Labs will carry enormous expectations. Success could re-establish Meta as a serious contender in advanced AI development. Failure, however, may further cement the company’s position as a follower rather than a leader in the AI race.
Meta’s Superintelligence Lab (MSL) is a new AI research division focused on developing advanced, general-purpose AI systems capable of reasoning, planning, and acting autonomously.
The lab is led by Alexandr Wang, co-founder of Scale AI.
“Mango” is Meta’s internal codename for a new image and video AI model, while “Avocado” refers to a text-based model aimed at improving coding and reasoning capabilities.
Meta plans to release both models in the first half of 2026.
Meta has faced internal restructurings, leadership changes, and talent churn, while competitors have moved faster in deploying high-performing AI models and products.
Not yet. Meta AI’s reach is largely driven by its integration into Meta’s existing social platforms rather than strong independent adoption.
The success of Mango and Avocado will likely determine whether Meta can reclaim relevance and leadership in the rapidly evolving AI ecosystem.
With pressure mounting and expectations high, Meta’s next generation of AI models could define the company’s future in artificial intelligence.
