Key facts
- India aims to lead in foundational AI model development.
- Varya, India's first indigenous AI video model, has been launched.
- The Varya platform was developed by Avataar. ai with support from the IndiaAI Mission.
- Varya aims to make high-quality video generation more affordable and accessible.
- The platform reduces video generation steps from 50 to four.
- The IndiaAI Mission has supported the creation of 20 foundational AI models, with five released.
India is asserting its ambition to lead in the development of foundational artificial intelligence models, with a senior government official stating the nation will not be outpaced in this technological race. Electronics and IT Secretary S Krishnan emphasized this point during the launch of Varya, India's first indigenous AI video model developed by startup Avataar. ai with support from the IndiaAI Mission.
Krishnan highlighted that Varya serves as a direct response to skepticism about India's capabilities in advanced AI platforms, demonstrating that such efforts can succeed domestically. The platform aims to democratize high-quality video creation, making it more affordable, accessible, and relevant for the Indian market.
Avataar CEO and Co-Founder Sravanth Aluru explained that Varya allows users to generate edited videos from photos and raw footage by simply inputting desired output details. He claimed the platform is approximately 27 times faster and cheaper than existing leading models, reducing the video generation process from an estimated 50 steps to just 4. This efficiency, he believes, is crucial for unlocking India's productivity curve through affordable AI.
Aluru anticipates that Varya will not only serve existing markets but also create new ones by enabling individuals and businesses to produce videos that were previously too expensive or cumbersome. He also addressed concerns about job displacement, stating that AI tools should complement human ingenuity and that the focus remains on empowering creators' ideas, removing the tedious aspects of video production while preserving creative value.
Under the IndiaAI Mission, startups have created 20 foundational AI models, with five now released. These include BharatGen's multimodal large language model designed for 22 Indian languages and a 105 billion parameter model by startup Sarvam. The government supports these initiatives through research consortia, subsidized GPU access, and support for selected technology organizations.