Indian AI startup Sarvam AI has announced an accelerated timeline for its trillion-parameter model launch, now targeting completion within six months from the current timeline. The company, which has secured $234 million in the first close of its $300 million Series B round at a post-money valuation of $1.5 billion, is making significant progress on its ambitious AI development plans. As reported by Business Standard, the company has 500 enterprise clients and is actively expanding its global presence with plans to open an office in San Francisco to tap AI talent internationally.
Addressing the accelerated timeline, Pratyush Kumar, co-founder of Sarvam, has indicated the company is open to government ownership as India advances its sovereign AI ambitions. Speaking in an interaction with Avik Das, Kumar explained that the government's involvement could be beneficial for advancing India's AI capabilities. The company's stance reflects the broader trend of India's AI sovereignty vision, where domestic companies are exploring partnerships with the government to build indigenous AI infrastructure and reduce dependence on global models.
At the Epoch conference, Sarvam AI unveiled plans to build a one trillion-plus parameter foundation model, one of the most ambitious AI projects announced by an Indian startup. As reported by NDTV Profit, the company said the model is being developed from scratch in India and will target advanced enterprise use cases, including coding, cybersecurity, scientific research, and large-scale simulations. Pratyush Kumar, co-founder of Sarvam, announced at the company's launch event in Bengaluru that "We are very happy to announce that we are building a trillion-plus parameter model right here in India. We are building them from scratch to be competitive in coding, cybersecurity, simulation, science and more." The startup also introduced a new pricing strategy for its flagship Sarvam 105B model, claiming it delivers AI inference at 5.5 times lower cost than comparable global models.
Sarvam is retailing its flagship Sarvam 105B model at $0.80 per one million blended tokens, compared to $4.50 for OpenAI's GPT-5.4 Mini and $9 for Gemini 3.5 Flash. According to the firm, this makes their models 5.5 times cheaper than GPT-5.4 Mini and over 11 times more affordable than Gemini 3.5 Flash. The company claims its 105-billion-parameter model delivers performance comparable to much larger AI models at a significantly lower cost. As reported by NDTV Profit, Sarvam 105B performed on par with GLM 5.2, a 753-billion-parameter model, across enterprise benchmarks such as tool use, robustness, consistency and process execution, despite the wide gap in parameter size. The company stated that if you are building a voice product today, Sarvam is the cheapest, fastest and most scalable solution, with the model hosted in India and having commenced work on serving voice AI workloads, processing up to crores of minutes per day.
Indian AI startup Sarvam AI has secured $234 million in the first close of its $300 million Series B round at a post-money valuation of $1.5 billion. The funding round was led by HCLTech with participation from Bessemer Venture Partners, while existing backers Khosla Ventures and Peak XV Partners also invested in the company. As reported by Business Standard, the company has appointed Devendra Singh Chaplot as a hands-on advisor to support the development of next-generation frontier AI models. Chaplot, who has updated his LinkedIn profile to reflect this development, will be based in Palo Alto, California, marking a significant addition to Sarvam's global AI leadership team. The appointment was announced at Sarvam's first developer conference, Epoch 2026, in Bengaluru.
