India's MeitY Launches Varya AI Video Model: Here's What It Does & How To Use It

India has taken a significant step in its sovereign artificial intelligence journey with the launch of Varya, a distilled AI video model developed by Bengaluru-based startup Avataar AI under the India AI Mission. MeitY Secretary S Krishnan announced the launch, describing it as part of India's commitment to periodically releasing homegrown sovereign AI models, with a high-quality video model being the latest milestone.

What Is Varya AI?

Avataar AI has launched Varya, a distilled video model built to understand local context, such as identifying different festivals, food, and clothing. The model is designed with India's cultural specificities at its core, addressing a longstanding gap in global AI video tools that have frequently produced generic or stereotyped representations of Indian life.

Varya did not start from scratch. It was built using Wan 2.2, a publicly available video generation model released by Alibaba, and used a technique called distillation, essentially compressing the model's capabilities into a leaner, faster version optimised for Avataar's specific use cases. The result is a model that runs in four steps rather than Wan 2.2's 50, producing video ten times faster and at a fraction of the cost.

The price advantage: A 20x difference

The company plans to charge Rs. 0.48 ($0.005) per second of video on its hosted service, far cheaper than models like Veo, Kling, Luma, and Runway, which typically charge $0.10 or more per second. That is a roughly 20x price difference.

Using an NVIDIA H200 GPU, Varya can generate a 5-second 720p clip in 45 seconds, compared to 1,230 seconds for Wan 2.2.

Here's what's currently known about how to access and use Varya AI:

Try It Directly

Anyone can try Varya now on its website using text prompts or reference images. The live demo is available at varya.avataar.ai — you can enter a text description or upload a reference image to generate a video clip.

Two Input Methods

Based on what Avataar has shared, you can generate videos in two ways: by typing a text prompt describing the scene or action you want, or by uploading a reference image as a visual starting point. The model is tuned to understand Indian cultural context, so prompts referencing Indian festivals, food, clothing, or architecture should yield more accurate results than with global models.

For Developers — Open-Weight Access

Avataar plans to release Varya as an open-weight model on India's AI Kosh portal (aikosh.indiaai.gov.in), along with its training data, so developers can self-host or modify it. This means developers will be able to download the model weights, run it on their own infrastructure, or fine-tune it for specific applications.

For Enterprises

The model is specifically targeted at applications in education, commerce, governance, citizen services, and digital storytelling for MSMEs. Avataar plans to make Varya available to enterprise customers and is also open to partnerships with video platforms like Higgsfield and Adobe Firefly.

Pricing

The company plans to charge Rs. 0.48 ($0.005) per second of video on its hosted service, making it significantly cheaper than most global alternatives.

The full developer API and AI Kosh portal listing are still being rolled out. The quickest way to start right now is the demo at varya.avataar.ai. You can enter a text description or upload a reference image to generate a video clip.

Based on what Avataar has shared, you can generate videos in two ways: by typing a text prompt describing the scene or action you want, or by uploading a reference image as a visual starting point. The model is tuned to understand Indian cultural context, so prompts referencing Indian festivals, food, clothing, or architecture should yield more accurate results than with global models.

Avataar plans to release Varya as an open-weight model on India's AI Kosh portal (aikosh.indiaai.gov.in), along with its training data, so developers can self-host or modify it. This means developers will be able to download the model weights, run it on their own infrastructure, or fine-tune it for specific applications.

The model is specifically targeted at applications in education, commerce, governance, citizen services, and digital storytelling for MSMEs. Avataar plans to make Varya available to enterprise customers and is also open to partnerships with video platforms like Higgsfield and Adobe Firefly.

Part of India's $1.2 billion AI Mission

The India AI Mission is a roughly $1.2 billion initiative that, among other things, gives selected startups access to subsidised GPU compute in exchange for releasing their models publicly. Avataar AI is one of 12 startups selected for the programme.

The government has also scaled its compute stack aggressively. Earlier this year, IT Minister Ashwini Vaishnaw said India aims to attract $200 billion in AI investment by 2028 and more than double its GPU capacity within six months.

MeitY Secretary S. Krishnan has indicated that India intends to keep pace with this momentum, with the government planning to release sovereign AI models periodically across different modalities as part of its broader digital self-reliance agenda.

Published on: Friday, June 12, 2026, 01:58 PM IST