India’s artificial intelligence ecosystem has long trailed the output of the United States, Europe, and China, with only a handful of startups releasing foundational models — and most of those have been large language or voice models. To accelerate development, the Indian government launched the India AI Mission, a roughly $1.2 billion initiative that provides selected startups with subsidized GPU compute in exchange for making their models publicly available. One of the 12 startups chosen for the program, Peak XV-backed Avataar AI, has now released Varya, a video generation model designed specifically to understand local Indian cultural context — including festivals, food, and clothing — while delivering inference at a fraction of the cost and time of existing alternatives.
How Varya Achieves 10x Faster Video Generation
Avataar AI, which builds video tools for e-commerce, did not develop Varya from the ground up. The team started with Wan 2.2, a publicly available video generation model released by Alibaba, and applied a technique called distillation — compressing the larger model’s capabilities into a leaner, faster version optimized for Avataar’s specific use cases. The result is a model that generates video in just four inference steps compared to Wan 2.2’s 50, producing output approximately 10 times faster and at a dramatically lower computational cost.
To put that performance in concrete terms: on an NVIDIA H200 GPU, Varya can generate a 5-second 720p clip in 45 seconds. Wan 2.2, by contrast, requires 1,230 seconds — over 20 minutes — to produce the same output. That speed advantage positions Varya as a practical option for real-time or near-real-time video generation at scale.
Pricing That Targets Population-Scale Adoption
Perhaps the most striking aspect of Varya is its pricing model. The company plans to charge ₹0.48 (approximately $0.005) per second of video on its hosted service. That is roughly 20 times cheaper than competing models such as Veo, Kling, Luma, and Runway, which typically charge $0.10 or more per second. The aggressive pricing reflects a deliberate strategy to make AI video generation accessible across India’s vast and diverse user base.
“India is a video-first market. We see this across every large consumer internet product in India: video wins over text. Current AI video models are too expensive for population-scale use in India. If video AI is going to reach students, teachers, MSMEs, creators, enterprises, and public services, costs have to come down dramatically. Cost is the biggest unlock for AI adoption in India,” Peak XV’s managing director Rajan Anandan told TechCrunch.
Training for Cultural Nuance
Image and video generation models frequently miss cultural nuances and produce stereotyped or generic outputs — a well-documented problem in AI. Avataar says it has used curated data to train Varya to recognize and accurately depict Indian cultural specifics including food, clothing, architecture, and festivals. This makes the model more useful for local creators, educators, marketers, and enterprises who need culturally appropriate visuals rather than the generic outputs typical of Western-trained models.
Open Access and the Broader Indian AI Push
Varya will be released as an open-weight model on India’s AI Kosh portal — the government’s centralized repository for publicly available AI models and datasets — along with its training data. This means developers can self-host or modify the model for their own needs. Avataar also plans to make Varya available to its enterprise customers and has signaled openness to partnerships with video tools including Higgsfield and Adobe Firefly. Anyone can try the model now on the Avataar website using text prompts or reference images.
Varya’s launch reflects a pragmatic tradeoff in India’s AI ambitions. Industry observers have noted that India can make its mark in AI by creating applications and fostering a robust developer ecosystem rather than competing on foundational model research. Model development has been slower in India due to a lack of compute and limited availability of high-quality training data. The India AI Mission — which selected 12 startups including Avataar last year — is part of a broader government push to close that gap. Earlier this year, IT minister Ashwini Vaishnaw stated that India aims to attract $200 billion in AI infrastructure investment by 2028 and plans to more than double its GPU capacity within six months.
What This Means for Developers and Creators
For developers, creators, and enterprises in India — and for anyone building applications that require culturally aware, cost-effective video generation — Varya represents a significant new option. The model’s open-weight release on the AI Kosh portal means it can be self-hosted, fine-tuned, and integrated into existing workflows without per-second API costs. Anyone interested can test Varya immediately via Avataar’s website using text prompts or reference images, making this a launch with immediate practical utility rather than a distant roadmap promise.
/i.s3.glbimg.com/v1/AUTH_08fbf48bc0524877943fe86e43087e7a/internal_photos/bs/2026/B/N/5EQ9QeQAa2XoA9mRtS0w/marvels-wolverine-state-of-play.jpg "Marvel’s Wolverine Settings: What Actually Works in Combat")
