On November 3, the IndiaAI Mission , under the Union Ministry of Electronics and IT, announced the release of three indigenous foundation language models. EKA-1 covers the 22 scheduled Indian languages and is designed for tasks such as mathematics, coding, and reasoning. The other two language models, Dhīr 2B and Dhīr 5B, support 10 Indian languages for uses like retrieving information from documents and supporting government and business services.
This is just the latest in a series of indigenous multilingual large language models (LLMs) that aim to bring Indian regional languages into mainstream artificial intelligence (AI) adoption. These models had subsidised access to national AI compute infrastructure and were built as part of the IndiaAI Innovation Centre pillar, set up in 2024, which enables developing indigenous and homegrown LLMs and domain-specific foundational models.
Computing power, however, is only one side of the problem in building for India’s diverse demography.…