Although India remains a major AI user, its homegrown models lag behind global leaders in development speed and capability, highlighting challenges and opportunities in building domestic AI expertise.
India is one of the world’s most active users of artificial intelligence, but it remains a small player in building the most capable homegrown models. Bloomberg Businessweek data show that between January and July 2026, Indian developers routed 94% of the AI work they sent through OpenRouter to models built in the United States or China.
That imbalance is echoed in Artificial Analysis’ tracking of 610 AI models. Of the 257 models currently available for sale, only two were built in India, both from Bengaluru-based Sarvam. On its 20 August 2026 data set, Artificial Analysis ranked Sarvam 105B as India’s strongest model, with a score of 11.9 out of 100, far below Anthropic’s Claude Opus 5 at 63.1 and China’s Kimi K3 at 59.7.
The gap is also visible in the pace of model development. Artificial Analysis says 171 of the 257 available models score above Sarvam 105B, while January alone saw 113 launches in the United States and 91 in China, compared with just two in India. OpenAI’s o1-preview had already exceeded Sarvam’s current score in September 2024, suggesting that India’s best publicly tracked model is still roughly where leading systems were two years ago.
Price, however, remains India’s clearest advantage. Sarvam 105B can deliver work that might cost $10 on Claude Opus 5 for about seven cents, and the model can also be downloaded and run for free. Sarvam has also positioned its systems around Indian languages such as Bhojpuri, Marathi and Tamil, a capability that is not reflected in the global benchmark tests used by Artificial Analysis.
The benchmarks themselves do not fully capture India’s broader role in AI adoption. Stanford’s AI Index showed that in 2025, 58% of workers worldwide were using AI regularly, while the figure in India and several other countries was above 80%. Reuters has previously reported that India is the second-largest market for Claude.ai usage after the United States, with demand heavily concentrated in software and web development.
That demand comes as the government tries to strengthen domestic capability through the IndiaAI Mission, including GPU access for local developers. According to reports from Kotak Neo and Moneycontrol, the state has also identified sovereign model proposals for support, while Sarvam has said it intends to build a trillion-parameter foundation model aimed at enterprise AI, coding, cybersecurity, scientific research and simulation. Even so, Nandan Nilekani has argued that India should not chase frontier models alone, and instead focus on becoming what he calls an AI “use-case capital”.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.
