In two separate developments, mobile unicorn Ola and artificial intelligence startup SarvamAI announced their Large Language Models (LLM) trained to generate Hindi text.

Sarvam released the first Hindi LLM OpenHathi-Hi-v0.1, which is built on Meta's Llama2-7B open source architecture and offers GPT-3.5 performance for Indian languages.

Besides Bhavish Aggarwal, founder, OLA took to X to announce the launch of its first LLM 'Krutrim' slated on December 15. Aggarwal dropped a video of Krutrim generating the invite for the launch event in both English and Hindi.

According to the company they are super excited to release OpenHathi-Hi-v0.1, the first Hindi LLM from our OpenHathi series of models. This model is trained under compute and data constraints to show that they can get GPT-3.5-like performance on Indic languages with a frugal budget.

According to the compay, they are planning to show model works as well as, if not better than GPT-3.5 on various Hindi tasks while maintaining its English performance. Along with standard NLG tasks, we also evaluate on a bunch of non-academic, real-world tasks.

Led by UIDAI veteran Vivek Raghavan and Pratyush Kumar, the 5-month-old Bengaluru-based startup raised $41 million this month. The round was led by Lightspeed Ventures, Peak XV Partners and Khosla Ventures, who are also investors in OpenAI.

Indian startups have established multilingual LLMs trained in Indian languages, including Corover.ai's recently launched BharatGPT and Socket Labs' upcoming Pragna model.

OLA’s Aggarwal says, “Very excited to share what Krutrim has been working on! India’s first AI. Full stack AI tech made in India. AI will transform everything, touch our economic, cultural lives so deeply. This time instead of using western products, India will build our own!”