Published Feb 19, 2026 | 2:04 PM ⚊ Updated Feb 19, 2026 | 2:04 PM
Sarvam AI.
Synopsis: Bengaluru-based startup Sarvam AI introduced two indigenous large language models trained specifically for Indian languages during the ongoing India AI Impact Summit 2026 in New Delhi. According to the company, the AI models are trained using 22 Indian languages on high-quality datasets.
Bengaluru-based startup Sarvam AI introduced two indigenous large language models (LLMs) trained specifically for Indian languages — a 30-billion-parameter model and a larger 105-billion-parameter model — during the ongoing India AI Impact Summit 2026 in New Delhi.
Sarvam Co-founder and CEO Dr Pratyush Kumar said the company’s 105-billion-parameter model performs well on most benchmarks and explained that although it is only about one-sixth the size of the 600-billion-parameter DeepSeek R1 model released last year, it was trained from scratch and delivers similar competitive intelligence.
“Sarvam Vision achieves state-of-the-art accuracy of 84.3 percent on the olmOCR-Bench (English only subset) outperforming frontier models like Gemini 3 Pro and recent OCR models like DeepSeek OCR 2,” Kumar said.
At the event, Sarvam also showcased its chatbot Vikram, which supports several Indian languages and English. The company said the name honours Indian physicist Vikram Sarabhai.
Sarvam was founded by Dr Vivek Raghavan and Kumar in August 2023.
“We want India to embrace the most important technological shift of our time with confidence and control. Our ambition is to build foundational components and apply them to the country’s unique needs. To this end, we’ve built a full-stack AI platform, with everything developed, deployed, and governed entirely in India,” the company said on its website.
“At Sarvam, we are committed to building India’s full-stack sovereign AI platform – from models grounded in Indian languages and datasets to applications deployed at population scale. We are grateful for these partnerships to provide us the opportunity to create impact at scale. Together we will build technology that reflects how India thinks, speaks, and solves its hardest problems. The future is intelligent, sovereign, and built locally,” Kumar said.
According to the company, the AI models are trained using 22 Indian languages on high-quality datasets, including varied financial documents, literature, newspapers, historic texts, and more.
Kumar said the platform supports AI video dubbing, and Sarvam Studio generates high-fidelity dubs in 11 Indian languages. In an expert study, participants preferred Sarvam Studio for overall quality and production readiness, he said in a social media post.
He said Studio excels in contextually translating long-form content across genres and that their evaluations demonstrated that readers strongly preferred the output from Studio across different genres.
The company had also launched Sarvam Kaze, a smart glass.”Sarvam Kaze moves intelligence from the screen to the real world. You wear it. It listens, understands, responds, and captures what you see. And you can build custom experiences for it with the Sarvam platform. This is a whole new world to build for,” Kumar said.
