Varya Video generation model
Why in News?
Varya is India’s first indigenous distilled video AI model. On June 11, 2026 Developed by Bengaluru-based startup Avataar AI with strategic backing from the government's India AI Mission, the model is designed to democratise high-quality video generation at a fraction of global costs.
Key Capabilities & Technical Specifications
- Core Architecture: Varya is a 14-billion-parameter AI model built using model distillation to compress the capabilities of Alibaba’s open foundational video model, Wan 2.2, into a lightweight version.
- Extreme Speed Leap: Through advanced distillation, it cuts video generation complexity down from 50 steps to just 4 steps.
- Processing Performance: Operating on an NVIDIA H200 GPU, Varya generates a 5-second 720p clip in approximately 45 seconds, compared to a staggering 1,230 seconds required by its base model.
- Hyper-Affordable Pricing: It slashes creation costs down to roughly ₹0.48 to ₹0.50 per second of video. This makes it up to 10 to 20 times cheaper than international alternatives like Gemini Veo or OpenAI Sora.
- Input Architecture: The interface works on a simple sequence flow where users can input text prompts or upload static images to seamlessly generate and extend active video scenes.
Hyper-Local Contextual Awareness
- Cultural Nuances: Unlike global models trained broadly on Western internet data, Varya understands India's specific regional environments, architecture, and daily public spaces.
- Social Visuals: It is finely tuned to accurately depict Indian clothing styles, community dynamics, traditional foods, and local festive settings without artificial distortions.
Population-Scale Deployments
- Education: Allows village teachers to instantly transform simple ideas into engaging visual lessons for local classrooms.
- MSMEs & E-Commerce: Enables small business owners to generate high-quality product advertisements and marketing tools without expensive camera crews.
- Public Information: Provides government agencies with a tool to rapidly roll out public service communications and visual stories to the masses.
