Varya: Here Is Avataar AI’s Highly Efficient and Culturally Aware Video Model Developed Under IndiaAI Mission

Avataar AI has unveiled Varya, a ground-breaking distilled video generation model designed to provide high-quality AI video tools tailored specifically for the Indian market. Developed with support from the Indian government’s IndiaAI Mission, Varya represents a significant advancement in indigenous AI innovation, aiming to make frontier-level video generation affordable and accessible for population-scale use cases.

Unlike conventional video generation models that require intensive computational power, Varya utilises a technique known as “model distillation.” By compressing the capabilities of a larger "teacher" model, specifically the publicly available Wan 2.2 model, into a leaner "student" version, Avataar AI has successfully reduced the generation process from 50 steps to just four. WhatsApp Football Emoji Update: Platform Introduces New Emoji to Trionda Match Ball; Check Details.

This breakthrough yields significant performance gains, allowing Varya to produce video ten times faster than its predecessor while maintaining comparable quality. In practical terms, the model can generate a 5-second 720p clip in approximately 45 seconds on an NVIDIA H200 GPU, a task that would otherwise take over 20 minutes.

A primary focus of the Varya model is its cultural awareness. While many global models produce generic or stereotyped visuals, Avataar AI has trained Varya on curated datasets to accurately recognise and replicate local nuances, including specific Indian clothing, architecture, festivals, and culinary traditions. This focus on local context is expected to be a major advantage for e-commerce, education, and digital content creation within the country.

Affordability is equally central to the launch. Avataar plans to charge approximately ₹0.48 per second of video, a price point nearly 20 times cheaper than leading international services. This aggressive pricing strategy is intended to ensure that AI-driven video tools can reach a wide range of users, including students, teachers, MSMEs, and public service providers, for whom current costs remain prohibitive.

The launch of Varya is a testament to the IndiaAI Mission, a government initiative with an outlay of approximately USD 1.2 billion. This mission is focused on democratising high-performance compute resources by providing startups and researchers with subsidised access to GPU infrastructure. Through the mission, the government has onboarded over 38,000 GPUs to help Indian startups develop foundation models that are multilingual and contextually relevant to the nation's unique socio-economic landscape. OpenAI GPT-5.6 AI Model Expected to Launch This Month Amidst Major ChatGPT Overhaul and Upcoming IPO Plans.

Varya is set to be released as an open-weight model on the IndiaAI Kosh portal, the government’s centralised repository for AI datasets and models. By making both the model and its training data public, Avataar AI aims to foster an ecosystem where developers can self-host or fine-tune the technology to meet specific needs, further reinforcing India's trajectory toward technological self-reliance in the global AI race.

(The above story first appeared on LatestLY on Jun 12, 2026 12:39 PM IST. For more news and updates on politics, world, sports, entertainment and lifestyle, log on to our website latestly.com).