Avataar launches Varya, offers ₹0.50-per-second AI video

Avataar.ai announced the launch of Varya, a 14-billion-parameter video generation model described by the company as a distilled, India-built model, per CNBC-TV18. CNBC-TV18 reports cofounder Sravanth Aluru said the model can generate video at approximately ₹0.50 per second and compared Varya's quality and cost to Alibaba's Wan 2.2, saying it is nearly 27 times cheaper. CNBC-TV18 and Inc42 note that the startup is backed by Peak XV and that Avataar has received support under the India AI Mission.

Per the published coverage, the concrete technical disclosures are limited to model scale and the company description of Varya as a distilled model; neither source provides an architecture-level paper or benchmark traces for independent verification. CNBC-TV18 reports cost-per-second comparisons with global models, stating Varya produces about 211 seconds of video for ₹100, versus roughly 3 seconds for Google Veo 3.1 and around 7 seconds for ByteDance's Seedance 2.0, numbers reported by CNBC-TV18 and presented by the company.

Industry observers have framed the current AI video market as dominated by large, compute-heavy projects from major cloud and platform companies. Inc42 reports that Avataar is betting on an efficiency-first approach rather than scale alone. Editorial analysis: companies pursuing efficiency over extreme scale often aim to reduce inference cost and broaden addressable users among creators and small businesses; historically, affordability gains matter most when quality and latency are competitive and when tooling and integrations are available.

For ML engineers and product teams, the main practical questions are reproducible quality, latency, and end-to-end production costs once orchestration, storage, and delivery are included. Editorial analysis: practitioners evaluating Varya should treat the reported ₹0.50 per second figure as a vendor claim until independent benchmarks compare output fidelity, resolution, runtime, and GPU/CPU requirements against alternatives.

For practitioners and observers: independent benchmark studies on visual fidelity and temporal coherence; published inference-cost breakdowns including serving infrastructure and postprocessing; publisher or creator case studies showing time-to-create and retention metrics; any release of a technical report or model card from Avataar; and uptake signals from education, commerce, and creator-platform partners. Industry context: broader acceptance will depend on integrations with existing video tooling and verifiable priors on hallucination, safety, and content moderation.