Meta Platforms is developing a powerful new AI image and video generation model, code-named Mango, as part of its growing investment in multimodal artificial intelligence. According to reports from The Wall Street Journal, the model is expected to launch in the first half of 2026, marking a major step forward in Meta’s race to compete with OpenAI, Google, and Anthropic.
During an internal Q&A session, Alexandr Wang, Meta’s new Chief AI Officer, and Chris Cox, Chief Product Officer, shared details about Mango’s development and the company’s broader AI roadmap. Following the news, Meta’s stock rose 2.3%, reflecting investor optimism about its next-generation AI plans.
Strategic AI Overhaul Inside Meta
Mango is part of a larger AI restructuring effort within Meta, led by Alexandr Wang, founder of Scale AI. After joining Meta in mid-2025, Wang assumed leadership of Meta Superintelligence Labs, a new division focused on building advanced generative and multimodal AI systems that CEO Mark Zuckerberg calls “critical to Meta’s future.”
The overhaul follows internal challenges with Llama 4, Meta’s previous language model, which underperformed expectations upon release. In response, Zuckerberg initiated an aggressive recruitment campaign, offering multi-million-dollar incentives to attract top-tier AI talent.
Meta’s current AI lineup already includes Movie Gen, a powerful model that creates videos, images, and even audio from text prompts, and the Llama series, which powers Meta AI — the assistant now integrated across Facebook, Instagram, and WhatsApp.
Shifting Priorities: From Metaverse to AI
Meta’s renewed AI focus coincides with a major shift away from metaverse development. Executives are reportedly planning to reduce the metaverse division’s budget by up to 30% in 2026, with potential layoffs expected early in the year, according to Bloomberg.
Instead, Meta is committing $70–72 billion in AI infrastructure spending for 2025, investing heavily in GPU clusters, data centers, and custom AI silicon to power its models. Despite investor concerns about rising costs, the company intends to further increase AI spending through 2026 as it positions itself for dominance in the generative AI market.
Meta’s New Direction: Proprietary Over Open Source
One of the most notable strategic shifts under Wang’s leadership is Meta’s move away from open-source AI. Future models like Mango and Avocado (Meta’s upcoming large language model) will likely adopt a more closed and monetizable approach — signaling Meta’s intent to directly commercialize its AI capabilities.
This transition mirrors a broader industry-wide pivot as AI companies balance innovation, competition, and commercial sustainability. For Meta, the Mango project represents not just another AI experiment — but a defining moment in its transformation from a social media giant to a full-scale AI powerhouse.
