Bengaluru-based AI startup Sarvam AI has launched artificial-intelligence models specifically designed for India, positioning itself as a domestic alternative to global tools like ChatGPT and Claude. The unveiling, held at the India AI Impact Summit 2026 in New Delhi, highlights the country’s push toward a “sovereign AI” ecosystem and greater self-reliance in emerging technologies.
The startup introduced two new models capable of operating via voice commands and supporting 22 Indian languages a key advantage in a nation of 1.45 billion people, many of whom are not proficient in English. “Today we show we can bring our own AI to a billion Indians,” said Sarvam co-founder Pratyush Kumar at the event.
Also read: Ixigo Expands into Europe with ₹125 Crore Acquisition of Spanish Rail Platform Trenes
Agentic AI for Enterprise Automation
Sarvam’s offerings also include Agentic AI models that can autonomously execute tasks such as coding, meeting scheduling, and other workplace functions with minimal human intervention. The company believes these AI agents could significantly accelerate enterprise automation in one of the world’s fastest-growing economies.
The launch of India-specific AI models trained from scratch comes amid a global AI race dominated by the United States and China. The Indian government is actively funding AI initiatives and incentivizing local model development to ensure the country, home to a vast pool of technical talent, remains competitive on the world stage.
Competing with Global Giants
Despite its ambitions, Sarvam faces steep competition from international AI leaders. The startup has raised over $50 million in funding from investors including Lightspeed Ventures LLC and Khosla Ventures, with a valuation of roughly $200 million. This is modest compared to OpenAI ($500 billion) and Anthropic PBC ($380 billion), or even European peers like Mistral AI ($13.25 billion), which is expanding in India with local language support.
Sarvam emphasizes its “India-first” approach, highlighting data security by running models entirely within the country. The models are trained on trillions of Indian datasets, particularly in native languages, enabling real-time deployment not only in regional languages but also mixed-language contexts such as Hinglish.
Superior Performance in Local Language Tasks
In benchmark testing, Sarvam’s AI models demonstrated high accuracy on tasks including optical character recognition for Indian scripts. The Sarvam vision model achieved over 84% accuracy on document intelligence tasks, outperforming global models many times its size. “Sovereignty matters much more in AI than building the biggest models,” said co-founder Vivek Raghavan.
India’s Vision for a Sovereign AI Ecosystem
The India AI Impact Summit 2026 has brought together global CEOs, AI founders, policymakers, and researchers as the country seeks to position itself as an alternative to US and Chinese dominance. By emphasizing cost-efficiency, language diversity, and local data sovereignty, India aims to lead in “democratizing AI.”
The launch of Sarvam’s models marks a significant step toward developing a homegrown AI ecosystem. Other Indian startups, such as BharatGen, have also introduced India-made models this week, aiming to export AI solutions to developing economies where US and Chinese platforms are less favored.