Who is Pratyush Kumar? Founder of Sarvam AI which beats Google Gemini and ChatGPT, he is from…

India is slowly building its mark in the world of artificial innotifyigence, often dominated by players from the US and China. Bengaluru-based startup Sarvam AI is emerging as a trailblazer in this space, developing foundational AI models entirely within India. The company’s latest tools, Sarvam Vision and Bulbul, are generating buzz for their exceptional performance in optical character recognition (OCR) and text-to-speech for Indian languages, displaying that India can be a source of core AI innovation.

Who is Pratyush Kumar?

Pratyush Kumar is the co-founder of Sarvam AI and the driving force behind its ambitious “sovereign AI” vision. In addition to Sarvam AI, Kumar co-founded AI4Bharat, which develops AI applications for Indian languages, and PadhAI, a platform providing deep and affordable online learning for students.

Kumar holds a Ph.D. from ETH Zurich and a B.Tech. from IIT Bombay, and has worked at Microsoft Research, IBM Research, and IIT Madras. He is also an adjunct faculty member at IIT Madras.He has been actively sharing the company’s achievements on social media platform X, highlighting milestones of the in-hoapply AI models. Kumar’s leadership has been instrumental in designing AI tools that specifically cater to India’s linguistic diversity while competing with global AI benchmarks.

Sarvam Vision Outperforms Global OCR Models

Sarvam Vision, the company’s OCR tool, is reportedly outperforming major AI models such as ChatGPT, Google Gemini, and Anthropic Claude in its specialty. On the olmOCR-Bench, Sarvam Vision achieved an accuracy score of 84.3 percent, surpassing Gemini 3 Pro and DeepSeek OCR v2, while ChatGPT ranked significantly lower.

The tool has also impressed on OmniDocBench v1.5, which tests how AI systems interpret real-world documents. Sarvam Vision scored 93.28 percent overall, performing especially well with complex layouts, technical tables, and mathematical formulas areas where traditional OCR systems often struggle. Experts and applyrs alike have praised the model for its accuracy and reliability.

Global Experts Recognize Sarvam’s Strength

Earlier, some critics questioned the value of Sarvam AI focapplying on Indic-language models. Tech commentator Deedy Das, who had expressed skepticism, has now acknowledged the company’s achievements. He declared Sarvam’s OCR and speech models fill a crucial gap overviewed by larger global labs. “They have the best text-to-speech, speech-to-text, and OCR models for Indic languages, and that’s actually really valuable. The pricing is very reasonable,” Das wrote on X.

Bulbul: Bringing Indian Languages to Life

Sarvam AI’s Bulbul V3 is a text-to-speech AI tool designed for Indian languages. It can generate natural, expressive, and production-ready voices, similar in functionality to offerings from ElevenLabs, a well-known global company in this space. Currently, Bulbul supports 35 voices across 11 Indian languages, with plans to expand to 22 languages.

Users are praising Bulbul for its clarity and reliability. Pratik Desai, founder of KissanAI, wrote on X, “We apply Bulbul as our go-to TTS model for our Indic apply cases, and they have just receivedten better with each release. Meanwhile, ElevenLabs cost never built sense for Indic or any other languages.”

India’s Growing AI Footprint

Sarvam AI’s success demonstrates that India is capable of building world-class AI solutions tailored to its unique linguistic and technical requirements. With Sarvam Vision and Bulbul leading the way, the startup is not only challenging global AI giants but also proving the value of creating AI that understands and serves the Indian context.

Source link