Ola Krutrim launches AI lab with Rs 2,000 Cr investment
Krutrim, Ola's AI venture, will also commit an investment of Rs 10,000 crore by next year.
Krutrim, Ola's AI venture, has launched an AI lab with large language models (LLMs ) Krutrim-2 and Krutrim 1, multilingual vision language model Chitrarth-1; speech LLM Dhwani-1, Krutrim Translate for text-to-text translation, and evaluation platform BharatBench.
Founder Bhavish Aggarwal took to social media platform X to share the update. He also announced an investment of Rs 2,000 crore into Krutrim via equity and debt, and a commitment of Rs 10,000 crore by next year.
“Also announcing India’s first GB200 deployment in partnership with NVIDIA! Will be live by March and we will make it the largest supercomputer in India by end of year,” the X post read.
LLMs
Krutrim launched its first LLM Krutrim 1 in January last year and introduced Chitrarth-1 in April.
“While we’ve been working on AI for a year, today we’re releasing our work to the open-source community and also publishing a bunch of technical reports. Our focus is on developing AI for India - to make AI better on Indian languages, data scarcity, cultural context etc.,” said Aggarwal on X.
Krutrim-2 is designed specifically for Indic languages. The LLM features a 12 billion parameter architecture as opposed to Krutrim-1 which had 7 billion parameters.
The LLM also supports long-form generation, multi-turn conversations, and document translation, as per the report published by the company.
Chitrarth-1 is a multilingual Vision Language Model (VLM) that integrates a state-of-the-art multilingual LLM with a vision module.
The company said that it was trained on multilingual image-text data and works across 10 Indian languages—Hindi, Bengali, Telugu, Tamil, Marathi, Gujarati, Kannada, Malayalam, Odia, and Assamese, and English as well.
Meanwhile, Dhwani is an end-to-end trained speech LLM, powered by Krutrim-1 LLM. According to the company, this enables its LLM to directly understand speech without a separate speech-to-text automatic speech recognition model.
The company also open-sourced the speech-to-text translation capabilities of its model.
Dhwani supports translation between Indic Languages and English. The supported Indic languages are English, Bengali, Gujarati, Hindi, Kannada, Malayalam, Marathi, Tamil, and Telugu.
The Vyakyarth-1 is an Indic text embedding model that also operates in the ten Indic languages—Hindi, Bengali, Gujarati, Kannada, Malayalam, Marathi, Sanskrit, Tamil, and Telugu as well as English.
The company also launched BharatBench, a platform that evaluates and sets benchmarks. “Since there was no global benchmark for Indic performance, we’ve developed “BharatBench”,” Aggarwal added.
(Shradha Sharma, Founder and CEO of YourStory, is an independent director at Ola Electric.)
Edited by Kanishk Singh