India Advances AI Mission: Government to Announce Second Round of Beneficiaries

[Exclusive] Govt to Select BharatGen, Fractal in Phase 2 of IndiaAI Mission

IndiaAI Mission will announce this move next week in an event in New Delhi. BharatGen is likely to be selected for the second phase of the IndiaAI Mission, marking a significant boost for its efforts to build open-source foundational models for Indian developers and researchers, sources in the department of science and technology told AIM. According to sources, IndiaAI Mission will announce the move next week in an event in New Delhi, alongside the announcement of seven other firms, including Fractal, for building India's foundational AI models. In May, Mumbai-based Fractal launched its open-source LLM, Fathom-R1-14B. The company claimed that the model delivers mathematical reasoning performance that surpasses o1-mini and o3-mini, and approaches o4-mini levels, all at a post-training cost of just $499. The company is slated to go for an IPO later th

Government to declare Round 2 beneficiaries of Rs 1,500 crore AI mission next week - The Economic Times

As part of the next round of incentives for the artificial intelligence (AI) foundational models under the IndiaAI Mission, the government is likely to select Avataar.ai, TechMahindra, Fractal Analytics, Genloop, IntelliHealth, Shodha AI, and Zenteiq Aitech Innovations. ET had reported on Friday that BharatGen is one of the eight companies that has been shortlisted by the ministry of electronics and IT (MeitY). Union IT and electronics minister Ashwini Vaishnaw is expected to announce the beneficiaries selected in the second round of screening next week. Initial public offering (IPO)-bound company Fractal Analytics had submitted a proposal to the IndiaAI Mission to build India's first large reasoning model (LRM) at a total project cost of Rs 118.8 crore, ET had reported in March. The LRM series will include a small model (two to seven billion parameters), a medium model (20-32 billion parameters), and a large state-of-the-art (SOTA) model (70 billion parameters) with up to one trillion training tokens. In terms of size, this is much larger than OpenAI's o1 and o3 reasoning models. BharatGen, a consortium of IIT researchers anchored by IIT Bombay and supported by the Department of Science and Technology, is also expected to receive the nod in this round. In July, BharatGen released a 2.9 billion parameter bilingual LLM called Param 1. The open-source model was pretrained on five trillion tokens in English and Hindi. IT services major Tech Mahindra, which launched the Indus Project, has created a 1.2 billion parameter model trained in 37 Indian dialects. In January this year, MeitY had announced an incentive allocation of Rs 1,500 crore for entities and individuals who proposed to build AI models from the ground up. By February 15, it had received 67 applications from Indian and global startups and researchers, followed by 120 more applications in the following month. The total number of applications has now reached 500. In the first round, four startups -- Sarvam, Soket Labs, Gnani.ai, and Gan.ai -- were approved to receive subsidised GPU compute for indigenous AI models. Separately, the government had also empanelled 10 GPU-as-a-service providers to build a common compute facility, wherein GPU compute can be accessed at less than $1 per hour -- among the lowest rates globally.