The India AI Impact Summit, scheduled to begin on February 16 in New Delhi, will bring together stakeholders from India and abroad to discuss developments in artificial intelligence. On the domestic front, attention is expected to centre on 12 Indian startups selected under the IndiaAI Mission to build indigenous foundation models trained on Indian languages and datasets. These companies are working on large language models (LLMs) and multimodal systems designed to address local linguistic, sectoral and governance requirements. The startups involved are as follows:
Sarvam AI
Sarvam AI is developing large language models tailored for Indian languages and contexts. Its work spans multilingual reasoning, voice-based applications and productivity tools for enterprises and public services. Under the IndiaAI Mission, the company is building a sovereign foundation model capable of tasks such as text generation, translation and conversational interfaces across Indian languages.
Soket AI
Soket AI Labs is working on open-source, large-scale AI models optimised for India’s linguistic diversity. The company is focusing on multilingual and multimodal foundation models aimed at use in sectors such as defence, healthcare and education, with support for more than 22 Indian languages.
Gnani AI
Gnani AI is developing a voice-first foundation model focused on multilingual speech processing and real-time voice interactions. The model is designed for low-latency, speech-to-speech communication and is intended for applications in customer support, education, accessibility and public-facing systems.
Gan AI
Gan AI is building a multilingual model with a focus on advanced text-to-speech capabilities. The company aims to deliver voice synthesis quality comparable to leading global systems, supporting use-cases such as audiobooks, voice assistants, content creation and media localisation in Indian languages.
Avaatar AI
Avaatar AI is working on domain-specific AI avatars designed for Indian use-cases. These models can be fine-tuned for sectors such as agriculture, healthcare and governance, supporting applications ranging from crop advisories and patient assistance to public grievance and query response systems.
BharatGen (IIT Bombay-led consortium)
BharatGen is a consortium led by the Indian Institute of Technology Bombay that is developing multilingual and multimodal foundation models of varying scales. These models are intended for applications across agriculture, legal services, education, financial services and healthcare, using datasets that reflect Indian languages and cultural contexts.
Zenteiq
Zenteiq is developing a multimodal foundation model named BrahmAI, focused on engineering intelligence and scientific computing. The model is designed to support technical simulations, optimisation problems and scientific research, with applications in deep-tech and engineering-led innovation.
Gen Loop
Gen Loop is building a suite of lightweight language models supporting all 22 scheduled Indian languages. The portfolio includes a base model, an instruction-tuned model and a moderation model, aimed at efficiency and safety for use in education, social platforms and enterprise applications.
Intellihealth (NeuroDX)
Intellihealth, operating under NeuroDX, is developing an AI model for EEG signal analysis. The initiative focuses on early detection of neurological conditions and advancing research in affordable brain–computer interfaces, with potential applications in clinical diagnostics and healthcare delivery.
Shodh AI
Shodh AI is working on an AI system to support material discovery by integrating artificial intelligence into laboratory workflows. The model is intended to assist in identifying new materials, including advanced alloys and battery components, by automating experimental design and analysis.
Fractal Analytics
Fractal Analytics is building what it describes as India’s first large reasoning model. The system focuses on structured reasoning and domain expertise, particularly in science, technology, engineering and mathematics, with applications in medical diagnostics, analytics and complex problem-solving.
Tech Mahindra Maker’s Lab
Tech Mahindra’s Maker’s Lab is developing a foundation model optimised for Indic languages and dialects, alongside an agentic AI platform for government and enterprise use. The initiative aims to support translation, administrative automation and e-governance workflows.