A man-made intelligence (AI) unicorn was born in India too. The important character is Crewtrim, an AI startup that’s constructing India's first large-scale language model (LLM).

Tech Crunch reported on the twenty sixth (local time) that Crutrim succeeded in attracting $50 million (about 67 billion won) in funding with a company value of $1 billion (about 1.338 trillion won). This round was led by Matrix Partners India.

Krutrim, founded by CEO Bhavish Agarwal in April last yr, became not only the primary Indian AI startup but in addition the fastest startup in India to develop into a unicorn. CEO Argarwal founded and succeeded in Ola, a car-sharing platform, and Ola Electric, a two-wheeled EV vehicle company, and in addition ranked his third company, Crew Trim, as a unicorn.

Krutrim, which implies 'artificial' in Sanskrit, held an event last month and announced that it’s developing an open source large language model (LLM) trained with greater than 2 trillion tokens. The model is trained in 20 Indian languages ​​and may respond in 10 of them.

We plan to release the beta version of the chatbot next month and release the API to developers and firms. It’s also known that the corporate is preparing to develop a dedicated AI chip.

“India must construct its own AI, and Krutrim is committed to constructing India’s first complete AI computing stack,” said CEO Agarwal.

Despite being one in all the world's largest startup ecosystems, India has yet to supply notable ends in the LLM sector.

Nevertheless, Asia's first and second largest billionaires, Mukesh Ambani, Chairman of Reliance Industries, and Gautam Adani, Chairman of Adani Group, declared at the tip of last yr that they might concentrate on AI. Particularly, Chairman Ambani announced in September last yr that he would develop an LLM in partnership with NVIDIA.

As well as, AI startup Sarbam announced that it released India's first Hindi AI model as open source in December last yr. This model, based on Mehta's 'Rama 2' 7B, is claimed to have equivalent performance to 'GPT-3.5' for Indian languages.

Reporter Lim Da-jun ydj@aitimes.com