Indian AI startup Sarvam has announced plans to build a foundational AI model with more than one trillion parameters, becoming the largest AI project to date. The company said the model is being developed entirely in India and is intended for coding, cybersecurity, scientific research, simulations and other enterprise workloads. The announcement was made by co-founder Pratyush Kumar during the company’s launch event in Bengaluru.
Sarvam also introduced pricing for its 105-billion-parameter model, saying it charges $0.80 per one million blended tokens. The company said this is lower than OpenAI’s GPT-5.4 Mini at $4.50 and Google’s Gemini 3.5 Flash at $9 for the same token volume. Sarvam added that its models are hosted in India and currently process voice AI workloads amounting to crores of minutes every day.
BREAKING: 🇮🇳 India’s Sarvam announces it’s building a trillion-plus parameter AI model from scratch that’s 5 times cheaper than global rivals. pic.twitter.com/rbBvAiUCCl
— Crypto India (@CryptooIndia) July 30, 2026
Sarvam expands enterprise AI portfolio with Vision Edge
Alongside its language models, Sarvam announced the commercial launch of Sarvam Vision Edge, an AI platform built for enterprise document processing. The company also introduced Sarvam Vision 2.0, which improves optical character recognition for business documents.
Sarvam said the Odisha government is already using the technology to digitize and extract information from land records. The company believes document intelligence will become an important part of its enterprise AI business, particularly for organizations handling large volumes of records.
Why Sarvam is competing on price instead of model size
Sarvam’s announcement suggests that India’s AI companies are taking a different route from many global competitors. While several companies continue racing to build the largest frontier models, Sarvam is trying to convince businesses that lower operating costs can matter just as much as parameter counts. For companies running AI services every day, inference costs often become a larger expense than the initial deployment.
That pricing strategy could appeal to Indian businesses and government agencies that need locally hosted AI without paying premium prices for overseas infrastructure. If Sarvam can maintain competitive performance while keeping costs low, it may find demand among organizations that value predictable operating expenses, data residency and support for domestic deployments over having the largest publicly available model.
Elsewhere in India, Karnataka has announced plans to launch the country’s first government-backed Artificial Intelligence University, aimed at advancing AI research, education, and innovation.
Enjoyed this? Bookmark DeFi Planet, explore related topics, and follow us on Twitter, LinkedIn, Facebook, Instagram, Threads, and CoinMarketCap Community for seamless access to high-quality industry insights
Take control of your crypto portfolio with DEFI PLANET PRO, DeFi Planet’s suite of analytics tools.
