New Delhi: On Friday, the Indian AI company Krutrim achieved something big! It became the fastest unicorn in the country and also the first AI unicorn. What does that mean? Well, it closed its first round of funding where investors, including Matrix Partners India, put in $50 million. This investment valued Krutrim at a whopping $1 billion.


COMMERCIAL BREAK
SCROLL TO CONTINUE READING

In simpler terms, a lot of people believe in Krutrim's potential and give them a lot of money to help them grow. Let's have a look at the details of the company. (Also Read: Now You Can Choose Any Hospital For Treatment; Know All About Game-Changing Rule For Health Insurance)


The Birth Of Krutrim Si Designs


Bhavish Aggarwal and Tenneti joined hands to establish Krutrim Si Designs, a tech venture under the umbrella of ANI Technologies Limited, the parent company of Ola Cabs and Ola Electric. (Also Read: No Need For Thermometer! Now This Smartphone Can Measure Your Body Temperature)


What Is The Meaning Of Word 'Krutrim'?


Named after the Sanskrit word for 'artificial,' Krutrim is not just a buzzword; it's a large language model (LLM) that has undergone training on a staggering 2 trillion 'tokens' -- these are like building blocks of language used in everyday conversations.


Models Of Krutrim


Krutrim debuts with a base model, set to hit the market next month. But hold onto your seats because the advanced Krutrim Pro is scheduled for an early release next year, promising cutting-edge capabilities for problem-solving and task execution.


Language Diversity


Krutrim comprehends 20 Indian languages and can generate content in 10 of them, including Hindi, Kannada, Marathi, and Telugu. The team proudly asserts that Krutrim outshines even GPT-4 in supporting Indic languages.


How Krutrim Works?

Krutrim employs a custom tokenizer to interpret various languages and scripts, making it a versatile linguistic wizard. In head-to-head comparisons with other open-source LLMs trained with similar data volumes, Krutrim emerges victorious across a range of industry-standard benchmarks.