IIT Madras’ Bodhan AI, AI4Bharat launches four open models for Indian languages

Indian Institute of Technology (IIT) Madras-incubated Bodhan AI has launched four AI models designed to support speech, translation and text recognition in Indian languages. Developed in collaboration with AI4Bharat, the models will be offered as digital public goods, enabling organisations to build and adapt AI applications for India’s multilingual education ecosystem.

The four models provide capabilities in automatic speech recognition (ASR), text-to-speech (TTS), machine translation and optical character recognition (OCR). According to Bodhan AI, the models are an early output of the broader Bharat EduAI Stack, which is envisioned as sovereign digital public infrastructure for India's multilingual education ecosystem.

Also read IIT Delhi launches PG diploma in solar systems and management at Sonipat campus

Bodhan AI models target multilingual education

The ASR model supports 27 languages, while OCR supports 23 languages. Bodhan-Translate supports 22 languages and the TTS model supports 23 languages.

The ASR model converts spoken language into text, while TTS converts text into natural-sounding speech. OCR can recognise and extract text from documents and images, and Bodhan-Translate can translate content across Indian languages.

Bodhan AI said the models can help build applications for learning, tutoring, practice and assessment. For example, students could speak to an AI tutor in their preferred Indian language instead of typing in English. OCR could help educational applications read textbooks, worksheets and handwritten answers, while translation technology could make educational content available across Indian languages.

The models have been trained and optimised using NVIDIA Nemotron open models and libraries, including the NVIDIA NeMo framework for ASR, machine translation and OCR. Bodhan AI has also post-trained NVIDIA Nemotron 3.5 ASR to support Indian languages, including regional dialects and accents.

The models are served using NVIDIA TensorRT-LLM and vLLM inference microservices. NVIDIA and Bodhan AI are also collaborating on datasets, training methods and evaluations for future foundational models for Indian languages.

IIT Madras director calls models step towards sovereign AI

Congratulating Bodhan AI on the launch, V. Kamakoti, Director, IIT Madras, said India's AI development needs to focus on technology that understands the country's linguistic diversity.

Kamakoti said, “India’s AI journey cannot be built on technology alone. It must be built on technology that understands India. The launch of these voice and vision models is an important step towards creating sovereign digital public infrastructure for AI in education that can serve our linguistic diversity and enable innovation at scale.”

He added, “By making these capabilities available to the wider ecosystem, Bodhan AI is helping create the foundations for an education system where every student and teacher can access the benefits of AI, irrespective of the language they speak.”

Also read IIT Kanpur, ISKCON sign agreement for value-based education and student well-being

Bodhan AI to build shared AI infrastructure

Mitesh Khapra, Principal Investigator at Bodhan AI, said the organisation's approach was to build infrastructure that could be shared by the wider ecosystem.

Khapra said, “Bodhan AI aims to build with the ecosystem, not compete with it. We built these voice and vision models and made them accessible as digital public goods on a digital public infrastructure so that efforts across the country don't remain fragmented. Instead, there is one common layer that can power all Edu AI in India, without every institution having to duplicate the same work.”

He added, “For Bodhan AI, the ultimate purpose of these foundational models is to make AI more accessible to teachers and students, particularly those who interact with technology in Indian languages. By opening access to foundational capabilities, we seek to enable edtech companies, startups, researchers, universities, technology companies and government partners to develop their own applications for Indian users.”

AI4Bharat partners with Bodhan AI

AI4Bharat, which specialises in Indian-language AI, multilingual modelling and datasets, is partnering with Bodhan AI on the initiative. The organisation develops open-source datasets, tools, models and applications for Indian languages.

Balaraman Ravindran, Head of the Wadhwani School of Data Science and AI (WSAI), IIT Madras, said building AI for India requires foundational models that understand the diversity of Indian languages and contexts.

He said, “Building AI for India requires going beyond adapting existing models to Indian languages. It requires developing foundational capabilities that understand the richness and diversity of our languages, contexts and educational needs.”

Ravindran added, “The models being released by Bodhan AI represent an important foundation for this effort. By making these capabilities available through open-weight models and APIs, we are enabling researchers, startups, technology companies and education innovators to experiment, adapt and build applications that are relevant to India at scale.”

Student Tutor Bot launched for Classes 6 to 12

Alongside the four foundational models, Bodhan AI has also launched Student Tutor Bot and Teacher Assistant Bot.

The Student Tutor Bot is designed for students in Classes 6 to 12 and is based on NCERT and SCERT curricula. Students can ask questions through text or voice in 22 Indian languages. The tool uses textbook content to provide explanations, examples and assessments.

The Tutor Bot can guide students step by step through problems and includes an interactive digital canvas for diagrams, calculations and other work. Student responses can also be used to inform subsequent learning support.

Teacher Assistant Bot to help create classroom content

The Teacher Assistant Bot is designed as an AI workspace for teachers to plan, create, assign, assess and refine classroom work.

Teachers can generate lesson plans, worksheets, quizzes, homework and revision sheets by entering details such as grade, subject, topic, duration, difficulty and assessment requirements. They can also upload student work for evaluation using specific marking criteria.

Bodhan AI said AI-generated content will not automatically become a classroom decision. Teachers can review, edit, regenerate or discard the output, keeping them in control of the teaching process.

Also read ‘Have seen 8 suicides’: Why hundreds of IIT Delhi B.Tech students are protesting over Rishikesh Kumar's death

Bodhan AI focuses on data privacy

Bodhan AI said data privacy and responsible deployment are central to its approach. Its architecture will include data anonymisation protocols and compliance with applicable national education data frameworks.

The hosted API infrastructure is being developed around a sovereign deployment model, which Bodhan AI said would give institutions and ecosystem partners greater control over how AI capabilities are deployed.

The organisation said its broader objective is to develop AI infrastructure that is affordable and suited to India's regulatory and institutional environment, with the principle of “AI for India, governed in India".