Avataar AI launches Varya video model at $0.005 per second, 27x cheaper than global rivals

Avataar AI Introduces Varya Under IndiaAI Mission

Bangalore-based Avataar AI has launched Varya, marking a significant milestone as India's first indigenous AI video generation model developed under the government's IndiaAI Mission 1

1

. The open-weight video generation AI model costs just ₹0.48, or $0.005 per second—roughly 27 times cheaper than comparable models from Veo, Kling, Luma, and Runway, which typically charge $0.10 or more per second 2

2

. This dramatic price difference positions Varya as a tool built specifically for population-scale adoption in a market where affordability determines accessibility. CEO and co-founder Sravanth Aluru, a former Deutsche Bank investment banker and Microsoft alum, emphasized that users will be able to create a 211-second video for every ₹100 they spend on the platform 4

4

.

Distillation Technique Drives Speed and Cost Efficiency

The Peak XV-backed startup didn't build Varya from scratch. Instead, it started with Alibaba Wan 2.2, a publicly available video generation model, and applied a machine learning technique called distillation to compress the model's capabilities into a leaner, faster version 1

1

. This process reduces video generation from 50 steps to just four, producing video 10 times faster and at a fraction of the cost 3

3

. 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 1

1

. The firm developed the platform using 14 billion parameters, with all compute and engineering resources sourced from India 4

4

.

Culturally Aware AI Addresses Local Context Gap

What distinguishes Varya from global competitors is its cultural specificity. Image and video generation models often miss cultural nuances and produce stereotyped or generic outputs—a persistent problem with Western-trained models 1

1

. Avataar AI used curated data to train Varya to recognize and accurately render Indian clothing, food, architecture, festivals, and everyday settings 2

2

. Rather than retrofitting a Western-trained model, this culturally aware AI approach ensures outputs that resonate with Indian businesses, education systems, and public services. Peak XV's managing director Rajan Anandan told TechCrunch that India is a video-first market where video wins over text across every large consumer internet product, making cost the biggest unlock for AI adoption [1](https://techcrunch.com/2026/06/11/cheaper-faster-and-culturally-aware-avataars-video-ai-is-built for-indias-scale/).

Government Support Fuels Sovereign AI Models Development

The roughly $1.2 billion IndiaAI Mission gives selected startups access to subsidized GPU compute in exchange for releasing their models publicly 1

1

. Electronics and IT Secretary S Krishnan announced that startups supported by the mission have created 20 foundational AI models, with five now released . These include IIT Bombay-based BharatGen's multimodal large language model designed for 22 Indian languages, and a 105 billion parameter model by Sarvam 3

3

. Varya will be released on the AI Kosh portal, the government's centralized repository for publicly available AI models and datasets, allowing developers to self-host or modify it 1

1

. This reflects India's broader strategy to build sovereign AI models rather than renting Western infrastructure 2

2

.

Strategic Bet on Affordability Over Peak Performance

Avataar AI, which has raised $55 million from Peak XV Partners and Tiger Global, originally focused on creating video tools for e-commerce 2

2

. Varya represents its first foundation model and a strategic pivot toward building infrastructure that prioritizes scale and accessibility over competing with frontier models from ByteDance's Seedance, Kuaishou's Kling, or other Chinese and US alternatives on raw quality 2

2

. The company plans to make the model available to enterprise customers and is open to partnerships with video tools including Higgsfield and Adobe Firefly 1

1

. Anyone can try it now on the company's website using text prompts or reference images. 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 [1](https://techcrunch.com/2026/06/11/cheaper-faster-and-culturally-aware-avataars-video-ai-is-built for-indias-scale/). MeitY Secretary Krishnan stated that for those questioning whether India can succeed in model development, Varya demonstrates that India won't be second to anybody in such efforts 4

4

. The question now is whether a video model optimized for affordability and cultural relevance can gain adoption faster than technically superior but expensive Western alternatives in a market of 1.4 billion people where cost competitiveness matters more than peak performance.