Meta’s Mango and Avocado: Inside the $14 Billion Bet to Challenge Open

When Alexandr Wang walked into Meta’s headquarters earlier this year, he brought with him more than just the credentials of a successful founder. The 27-year-old entrepreneur, who had built Scale AI into a data-labeling powerhouse, was handed a mandate that would make most executives blanch: catch Meta up to OpenAI and Google in the AI race, and do it fast.

“The projects are being positioned internally as a major inflection point for Meta’s AI ambitions.” — Wall Street Journal

The $14 Billion Gambit

Meta’s acquisition of a near-majority stake in Scale AI, valued at more than $14 billion, wasn’t just about bringing Wang into the fold. It was a signal that Mark Zuckerberg was done playing catch-up. The company had watched as OpenAI’s GPT series and Google’s Gemini family captured the industry’s imagination while Meta’s Llama models, despite their open-source popularity, remained a step behind on the frontier.

Mango and Avocado represent Meta’s answer. According to sources familiar with the projects, Mango is being designed as an image- and video-generation model targeting high-quality creative output, positioning it directly against OpenAI’s Sora and Google’s Gemini 3 Flash. Avocado, meanwhile, aims to solve Meta’s weakest flank: advanced reasoning and coding capabilities where Llama has consistently lagged.

The timing matters. Meta’s AI efforts have been fragmented across multiple teams and priorities. Wang’s arrival brought a consolidation—the creation of Meta Superintelligence Labs, a tightly knit group of elite researchers and engineers given the resources and autonomy to move fast.

From Next-Word Prediction to World Models

What makes Avocado particularly intriguing isn’t just its ambition to match GPT-5.4 on coding benchmarks. Wang has reportedly described the model as part of a broader shift toward “world models”—AI systems that move beyond predicting the next token to actually understanding how the world works.

“If successful, this approach could mark a meaningful step toward more perceptive, context-aware AI.” — Industry Analyst

The technical bet is significant. Traditional large language models excel at pattern matching and text generation but struggle with causal reasoning and physical intuition. World models aim to bridge that gap, creating AI systems that can simulate consequences, plan multi-step actions, and operate with something closer to genuine understanding.

The competitive stakes couldn’t be higher. OpenAI’s GPT-5.4, released earlier this month, has set new benchmarks for reasoning and tool use. Google’s Gemini 3.1 Pro is gaining traction in enterprise deployments. If Meta can’t close the gap with Mango and Avocado, it risks becoming irrelevant in the AI infrastructure layer that increasingly defines the technology sector.

The Creative Generation Arms Race

Mango enters a market that’s heating up rapidly. OpenAI’s Sora demonstrated the potential of AI-generated video, but access remains limited. Google’s Veo 3 has shown impressive capabilities but hasn’t achieved broad deployment. The window for Meta to establish a foothold in AI-powered creative tools remains open—but it’s closing.

The model’s focus on “high-quality creative generation” suggests Meta is targeting professional use cases, not just consumer filters and effects. This aligns with the company’s broader push to become an infrastructure provider for the metaverse and beyond. If Mango can generate production-ready video content, it becomes a platform play, not just a feature.

“We’re past the hype cycle now. Companies that can demonstrate real value—measurable, repeatable, scalable value—are the ones that will define the next decade of AI.” — Venture Capital Partner

The Road Ahead

Industry observers expect Mango and Avocado to debut in the first half of 2026, potentially as early as Meta’s next developer conference. The launch will be scrutinized not just for raw capabilities but for how Meta positions the models in its broader ecosystem.

Will they be open-sourced like Llama, or kept proprietary to drive competitive differentiation? Will they be integrated into Meta’s existing products—Instagram, WhatsApp, Facebook—or offered as standalone developer platforms? The answers will reveal much about Zuckerberg’s strategic thinking.

For now, one thing is clear: Meta has made its move. With $14 billion and one of the industry’s brightest minds leading the charge, the company is betting big that Mango and Avocado can turn the tide. The rest of the industry is watching to see if this time, Meta can deliver.

This article was reported by the ArtificialDaily editorial team. For more information, visit MIT Sloan Management Review and The Wall Street Journal.