BENGALURU — February 2026 — Sarvam AI has introduced Sarvam Akshar, a new initiative focused on building high-quality language AI capabilities tailored specifically for India’s diverse linguistic landscape.

Sarvam Akshar is designed to strengthen foundational AI models across Indian languages, addressing long-standing gaps in speech recognition, text generation, translation, and understanding for regional scripts and dialects. The company said the goal is to make AI systems more inclusive and accessible for India’s multilingual population.

India has dozens of major languages and hundreds of dialects, many of which are underrepresented in global AI training datasets. Sarvam AI said Akshar aims to improve performance across these languages by investing in high-quality data curation, model training, and evaluation tailored to local linguistic nuances.

The initiative focuses on both text and speech capabilities, enabling more natural voice assistants, government services, customer support systems, and education tools in native languages. By improving language coverage and accuracy, Sarvam AI hopes to expand AI adoption beyond English-speaking users.

The company emphasized responsible development and collaboration with academic institutions, public bodies, and ecosystem partners to ensure linguistic diversity is respected and preserved.

Sarvam Akshar builds on India’s broader ambition to strengthen domestic AI capabilities, particularly in language technologies that reflect local culture, scripts, and communication patterns.

With the launch of Sarvam Akshar, Sarvam AI is positioning itself as a key player in India’s effort to create AI systems that speak, understand, and serve the country in its own languages.