India is actively engaged in the development and deployment of Artificial Intelligence (AI) models, with a strong focus on creating solutions tailored to its unique linguistic and cultural diversity, as well as addressing local needs in various sectors. The government's IndiaAI Mission, approved in March 2024 with a significant outlay of ₹10,300 crore over five years, is a key driver behind these efforts.
Here's a breakdown of prominent Indian AI models and initiatives, including those under development and those live for users:
Indian AI Models Under Development:
India is focusing on building its own foundational models, including Large Language Models (LLMs) and Small Language Models (SLMs), to reduce reliance on imported technology and ensure AI solutions are contextually relevant.
- BharatGen: This is the world's first government-funded multimodal LLM initiative, launched in 2024. It aims to enhance public service delivery and citizen engagement through foundational models in language, speech, and computer vision. BharatGen involves a consortium of AI researchers from premier academic institutions in India
and is anticipated for completion by 2026. - Sovereign LLM under IndiaAI Mission: The Government of India has selected SarvamAI to build the country's sovereign Large Language Model (LLM) under the IndiaAI Mission. This initiative marks the first time India will develop an indigenous foundational AI model from scratch, designed for voice and fluent in Indian languages. SarvamAI is developing three model variants:
- Sarvam-Large: For advanced reasoning and generation.
- Sarvam-Small: For real-time interactive applications.
- Sarvam-Edge: For compact on-device tasks. The goal is to build multi-modal, multi-scale foundation models from scratch, with initial versions expected to be ready within 4 to 10 months (as of January 2025).
- Indigenous GPU Capabilities: India aims to develop its own Graphics Processing Units (GPUs) within the next three to five years to further strengthen domestic capabilities and reduce reliance on imported technology.
- IndiaAI Dataset Platform (AI Kosh): This platform provides seamless access to high-quality, non-personal datasets to empower Indian startups and researchers to develop advanced AI applications. It will house the largest collection of anonymized data, ensuring diverse and abundant datasets for training models.
- AI Centres of Excellence (CoE): The government has established several AI CoEs in critical sectors:
- Healthcare
- Agriculture
- Sustainable Cities
- Education (a new CoE announced in Budget 2025 with an outlay of ₹500 crore).
- Call for Proposals for Foundational AI Models: As part of the IndiaAI Mission, the government has issued a call for proposals to support the development of foundational AI models using Indian datasets,
inviting researchers, startups, and entrepreneurs to collaborate.
Indian AI Models Live for Users:
Several Indian AI models and platforms are already live and making an impact across various sectors.
- Sarvam-M: This is a 24-billion-parameter hybrid open-weights large language model developed by SarvamAI, built on Mistral Small. It is praised for its strong performance in Indian languages, mathematics, and programming. Sarvam-M supports 10 Indian languages and is publicly accessible via Sarvam's API and on Hugging Face, encouraging developers to build, test, and contribute.
- Sarvam-1: Launched in October 2024, this is an open-source large language model optimized for Indian languages, supporting ten major Indian languages in addition to English. It is designed for applications such as language translation, text summarization, and content generation.
- Chitralekha: An open-source video transcreation platform developed by AI4Bhārat, Chitralekha enables users to generate and edit audio transcripts in various Indic languages.
Its functionalities include subtitle generation, audio/video dubbing, and video translation. - Hanooman's Everest 1.0: Developed by SML, Everest 1.0 is a versatile multilingual AI system that currently supports 35 Indian languages, with plans to expand to 90. Powered by the Executable Expert Model (EEM) architecture, it excels in real-time data access, predictive analytics, and image analysis tasks, enhancing accessibility in sectors like customer service, education, healthcare, and finance.
- Bhashini: Developed under India's Digital India initiative, Bhashini is designed to break language barriers by enabling real-time speech and text translation across multiple Indian languages.
It provides machine translation, speech recognition, and sentiment analysis, making AI more inclusive for diverse linguistic groups. Bhashini also makes its Indian language data available for startups. - BharatGPT (by CoRover.ai): This is India's own Generative AI platform, available across channels in 14+ Indian languages (voice) and 22+ languages (text). It is a human-centric conversational AI platform with contextual Generative AI capabilities, used by over a billion users through various virtual assistants for government and private organizations (e.g., IRCTC, LIC, NPCI). BharatGPT focuses on keeping data within India and is fine-tuned for Indian users.
- Niramai (Non-Invasive Risk Assessment with Machine Intelligence): This AI-powered thermal imaging solution is used for early detection of breast cancer, offering a cost-effective, non-invasive, and accessible screening method.
- Satyukt: This platform utilizes satellite data and AI to provide actionable insights for farmers, helping with crop monitoring, soil health analysis, and water management to improve agricultural productivity.
- NVIDIA Nemotron-4-Mini-Hindi-4B: This compact yet powerful Hindi language model, part of NVIDIA's NIM microservice, is designed to empower businesses to create AI solutions tailored to regional demands. Tech Mahindra has integrated it into its Indus 2.0 platform.
India's AI strategy emphasizes inclusivity, affordability, and innovation, aiming to leverage AI for societal and industrial advancements while ensuring ethical and responsible deployment. The country is also building significant computing infrastructure, including acquiring thousands of high-performance GPUs, to support its growing AI ecosystem.