Meta Superintelligence Labs
Meta's next-generation diffusion-transformer model that understands physics, causality, and temporal continuity.
Real-time generationPlans from $9.90/moUp to 10s video
500M+
Daily Video Viewers
Instagram data advantage
50+
Research Team
Engineers & AI specialists
10s
Video Length
High-fidelity output
DiT
Architecture
Diffusion Transformer
Get Inspired
Explore examples generated by Mango AI. Click to copy any prompt and try it yourself.
About
Mango is Meta's codename for a next-generation multimodal image and video generation AI model, developed inside Meta Superintelligence Labs (MSL) — Meta's elite AI research division led by Alexandr Wang. First revealed during an internal Q&A session on December 18, 2025, Mango represents Meta's most ambitious push into generative media.
Built on a diffusion-transformer (DiT) architecture, Mango goes beyond simple pixel generation — it learns the laws of physics alongside visual synthesis, enabling generated objects to maintain realistic shape, mass, and velocity over time. This "world model" approach dramatically reduces the physics hallucinations that plague other video generation models.
Capabilities
Everything you need to create professional-quality AI-generated media
Workflow
01
Enter a detailed text prompt. Mango understands complex compositions, styles, and cinematic directions.
02
The diffusion-transformer processes your prompt through its world model for physically accurate results.
03
Fine-tune with camera controls and style presets. Share directly or integrate with your workflow.
Architecture
The technology powering Meta's most advanced generative model
Comparison
See how Mango stacks up against leading AI video generation models
Use Cases
Roadmap
December 2025
Revealed during an internal Q&A led by Alexandr Wang and Chris Cox. The Wall Street Journal broke the story.
Early 2026
50+ engineers assembled, 20+ researchers recruited from OpenAI. $14.3B invested in Scale AI for data annotation.
Spring 2026
Limited access following the Llama-2 release strategy — select developers and partners first.
H1 2026
Full release with Instagram Reels and WhatsApp integration, reaching billions of users.
Pricing
All plans include access to our core features. Cancel anytime.
MonthlyAnnuallySave 50%
Basic
$9.90/mo
Billed $118.80/year
Save $120.00 / year
4,800 credits/year
Ideal for hobbyists and beginners
Standard
$19.90/mo
Billed $238.80/year
Save $240.00 / year
9,600 credits/year
Perfect for most creators
Pro
$49.90/mo
Billed $598.80/year
Save $600.00 / year
24,000 credits/year
Ideal for power users
Trusted by 1,000+ creators14-day money-back guarantee
FAQ
Mango AI is Meta's next-generation image and video generation model, built on a diffusion-transformer architecture with world-model physics understanding. It's being developed inside Meta Superintelligence Labs (MSL) and is expected to launch in the first half of 2026.
Mango AI's key differentiator is its world model approach — it learns physics, causality, and temporal continuity alongside visual generation. This reduces the physics hallucinations common in other models. It also benefits from Meta's massive Instagram data advantage and planned integration with Instagram Reels and WhatsApp.
Meta is targeting a first-half 2026 launch. Private beta invites are expected in spring 2026, following a strategy similar to the Llama-2 release. The public launch will include integration with Instagram Reels and WhatsApp.
Mango AI offers three subscription plans — Basic ($9.90/mo), Standard ($19.90/mo), and Pro ($49.90/mo). Each plan includes a monthly credit allocation that resets every billing cycle. You can save 50% by choosing annual billing. All plans include commercial use license, no watermarks, and private generation.
Mango AI can generate high-fidelity images from text prompts, 5-10 second video clips with temporal coherence, video-to-video transformations, and content with controllable camera motions, lighting, and style presets. It also features industry-leading lip-sync capabilities.
Mango AI is developed by Meta Superintelligence Labs, led by Alexandr Wang (the 28-year-old founder of Scale AI). The team includes 50+ engineers and AI specialists, with 20+ researchers recruited from OpenAI by Mark Zuckerberg personally.
Be among the first to experience Meta's revolutionary image and video generation model.
This platform is an independent product and is not affiliated with Meta Platforms, Inc.