Meta is preparing a major artificial intelligence push for 2026, aiming to reclaim lost ground in the rapidly intensifying AI arms race. According to reports from an internal Q&A session, Meta is developing two next-generation AI models code-named Mango and Avocado under its newly formed Meta Superintelligence Labs (MSL). These models are positioned as Meta’s first true “frontier AI” systems, designed to rival offerings from OpenAI and Google.
The initiative is being led by Alexandr Wang, Meta’s Chief AI Officer and founder of Scale AI, who joined the company following Meta’s reported $14 billion stake acquisition. Alongside Chief Product Officer Chris Cox, Wang outlined a bold shift in Meta’s AI philosophy moving beyond traditional language prediction toward building more advanced “world models.”
Mango: Meta’s Multimodal Creative Engine
Mango is Meta’s answer to OpenAI’s Sora and Google’s Gemini 3 Flash, particularly its viral “Nano Banana” image generation tool. Focused on high-quality image and video generation, Mango is expected to offer advanced multimodal reasoning, allowing it to interpret and understand visual data in a more human-like way.
Unlike current prompt-based generators, Mango aims to support professional-grade video production, social media content creation, and creative workflows. This positions Meta to better leverage its massive creator ecosystem across Instagram, Facebook, and Threads.
Avocado: A Leap in Text, Coding, and Reasoning
Avocado represents Meta’s most ambitious Large Language Model to date. Internally described as a “generational leap,” it is designed to close the gap with GPT and Gemini in coding, technical reasoning, and complex problem-solving areas where the Llama series has lagged.
In a significant strategic shift, Avocado may be released as a closed-source model, signaling Meta’s intent to monetize advanced AI capabilities rather than rely solely on open-source leadership.
Building World Models and a Super Team
Wang emphasized that Meta’s future lies in world models AI systems trained on massive amounts of visual data to understand and predict real-world behavior, not just generate text. To accelerate this vision, Mark Zuckerberg personally spearheaded the recruitment of over 20 top researchers from OpenAI, forming a “super team” of more than 50 elite scientists within MSL.
Why 2026 Matters
With Google’s Gemini reportedly surging past ChatGPT in app rankings and OpenAI dominating AI video generation, Meta faces intense competitive pressure. Mango and Avocado, both expected in the first half of 2026, represent Meta’s pivot from an open-source provider to a full-scale frontier AI platform.
If successful, 2026 could mark Meta’s return as a serious leader in next-generation artificial intelligence.
