Meta’s AI Ambitions Intensify with ’Mango’ Model as Rivalry with Google, OpenAI Escalates

Meta CEO Mark Zuckerberg unveiled plans for a new AI model codenamed Mango, targeting deployment by mid-2026 to compete with Google's Gemini and OpenAI's offerings. The announcement came during an internal Q&A with Chief AI Officer Alexandr Wang, who confirmed parallel development of 'world models'—AI systems that learn from environmental visual data.

The company has aggressively restructured its AI division under Meta Superintelligence Labs, poaching over 20 OpenAI researchers and assembling a 50-person team specializing in deep learning. The hiring spree reflects Meta's strategic pivot toward generative AI, particularly image synthesis, where it faces stiff competition from rivals.

September saw Meta release Vibes, a video generator developed with Midjourney, only for OpenAI to counter days later with Sora. This tit-for-tat innovation cycle mirrors earlier moves like Google's Nano Banana update, which propelled Gemini users from 450 million to 650 million between July and October.