
OpenAI has made Indic language performance and voice a top research priority in India, its global head of startups Marc Manara said. The firm slashed prices for its GPT-5.6 Luna and Terra…
OpenAI has made Indic language performance and voice a top research priority in India, its global head of startups Marc Manara said. The firm slashed prices for its GPT-5.6 Luna and Terra models by 80% and 20% last week to attract price-sensitive Indian startups. OpenAI’s startup team in India is now its largest in Asia-Pacific. The Indian conversational AI market was worth $653 million in 2025 and is expected to grow to $5.9 billion by 2034, according to IMARC Group.

Separately, Bengaluru-based Sarvam AI on July 31 launched Epoch Builder Edition, a platform for building Indian-language LLMs. It features 7-billion and 70-billion parameter models trained on 10 Indian languages, with tools for training, deployment and red-teaming. Sarvam AI has partnerships with three IITs and two state governments for pilot projects. The platform will be available in private preview from August 2026.
Two narratives are clashing: one that only Silicon Valley giants can solve India’s AI needs, and another that only Indian startups can get local languages right. In reality, OpenAI’s price cuts and Sarvam’s Epoch platform are complementary moves in a fast-growing market. The real test will be adoption by banks and government departments by late 2026. Can proprietary models keep up with open-weight alternatives on cost and accuracy for code-mixed Hinglish?
Sources (2): livemint.com, thehindu.com
This story was synthesised by AI from the 2 sources linked above.
Updated: this story now draws on 2 sources.