India's Avataar AI has introduced a video model that is priced at $0.005 per second, making it 27 times more affordable than competitors.

      Avataar AI has introduced Varya, an open-weight video model priced at $0.005 per second, making it 27 times more affordable than competitors. Developed under India’s AI Mission, it accurately represents Indian culture.

      Based in Bangalore, Avataar AI has launched Varya, among India’s first indigenous video AI models. It produces videos at approximately $0.005 per second, equivalent to 0.48 rupees. According to founder Sravanth Aluru, a former Deutsche Bank investment banker and an alum of Microsoft and IIT Mumbai, this cost is significantly lower than similar open-source video models.

      The cost efficiency is achieved through a process of distillation. Avataar started with Alibaba’s Wan 2.2, a publicly accessible video generation model, and streamlined it into a more efficient version that operates in four steps instead of the original fifty. This results in video generation that is ten times faster at a significantly lower price. Competing models like Veo, Kling, Luma, and Runway usually charge $0.10 or more per second.

      Varya does not aim to match the quality of advanced US and Chinese models. Products like ByteDance’s Seedance, Kuaishou’s Kling, and Alibaba’s Wan have taken motion realism and audio generation beyond what Varya provides. Instead, Varya focuses on scale and accessibility in a market of 1.4 billion people, where affordability is more critical than top performance.

      What sets Varya apart is its cultural specificity. Rather than adapting a model trained in the West, Avataar utilized curated data to train Varya, ensuring it accurately depicts Indian clothing, food, architecture, festivals, and daily life spaces. Global models that predominantly learn from Western datasets often produce outputs that misrepresent Indian culture, limiting their effectiveness for Indian businesses, education, and public services.

      The model will be open-weight and released on India’s AIKosh portal, the government’s centralized repository for AI models and datasets. Avataar is among twelve startups chosen for the IndiaAI Mission, a $1.2 billion initiative that provides selected companies with subsidized GPU computing in exchange for making their models publicly available.

      Avataar has secured $55 million in funding from Peak XV Partners and Tiger Global, originally focusing on developing video tools for e-commerce. Varya represents its first foundational model, aligning with a trend of Indian startups creating indigenous AI rather than relying on Western infrastructure. Earlier this year, Sarvam and BharatGen also launched their foundational models under the same program.

      India’s AI strategy differs from those of Europe and China. Instead of aiming to build the largest models, it seeks to develop models that effectively serve its population at a price point that the market can accept. Priced at $0.005 per second, Varya is exploring whether a video model designed for cost-effectiveness and cultural relevance can achieve faster adoption than a technologically superior yet pricier Western counterpart. In a nation where AI startups are already catering to local demands, the answer could very well be affirmative.