The ongoing discussion around AI models has typically revolved around American and Chinese innovations. However, Indian startup Sarvam AI is making headlines by developing foundational AI models that challenge even the most popular global contenders. Recently, their two innovative tools, Sarvam Vision and Bulbul, have gained considerable attention for their high performance in AI benchmarks.

Sarvam Vision, one of their flagship products, has particularly excelled in Optical Character Recognition (OCR), achieving remarkable accuracy ratings that surpass those of major AI models including ChatGPT, Google Gemini, and Anthropic Claude. This success has garnered praise not only from users but also from industry experts who are acknowledging the potential of homegrown AI solutions. Co-founder Pratyush Kumar has actively shared these achievements on social media, emphasizing the strong performance of their in-house models.

According to the information shared by Kumar, Sarvam Vision has achieved an impressive accuracy of 93.28% on the OmniDocBench v1.5 benchmark, specifically on the English-only subset of the dataset. This milestone underscores the effectiveness of their tools and positions Sarvam AI as a key player in the rapidly evolving AI landscape. As more Indian companies venture into AI development, Sarvam's achievements may signal a significant shift in the competitive dynamics of this technology sector.