Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
The founder of the builder Bhavish Aggarwal invests $ 230 million Founded AI Start As the country forced itself in an area where US and Chinese firms are dominated.
Aggarwal, in Krutrim, in a source with a source of investment, which is mainly explained by the Family Office, TechCrunch A post on x On Tuesday, Aggarwal, Krutrim said he tried to attract $ 1.15 billion after next year. The source said he would try to lift the rest of the capital from outside investors.
The financing announcement, unicorn Startup coincides with crutch, opening open sources and open sources and opening plans to create India’s largest supercomputer with NVIDIA.
This laboratory Krutrim-2, which is a parameter of 12 billion billion parameter, which is a strong performance in processing Indian languages, left Krutrim-2. Crutrim tests in sentiment analysis Shared Tuesday80% of the Code of the Code has reached 80% of success compared to 0.95 compared to 0.70 for competitive models.
Laboratory, images, speech translation and text searches are several specialized models, including several specialized models, including images optimized for Indian languages.
“We still approach global criteria together, but we did not make good progress for a year,” he said. Create a world-class Indian AI ecosystem. “
The initiative comes, as India tries to build itself in the artificial intelligence landscaped by US and Chinese companies. Last Release of DeepSeek’s R1 “Reasoning”It was built on a modest budget, sent shock waves through the technological industry.
India last week Praised DeepSEEK’s progress and the country’s Chinese servers will host large language models of the Chinese AI laboratory. Krutrim’s cloud arm Started offering DeepSEEK on Indian Servers last week.
My crouted also developed its own appraisal framework, BharatbenchTo perform the knowledge of AI models in Indian languages, primarily in the existing tendencies directly to English and Chinese.
The technical approach of the laboratory includes the use of 128,000 Token Context window, which allows systems to manage longer texts and more complex conversations. Performance measurements published by the start, grammar (0.98) and multi-turn conversations (0.91) showed Krutrim-2.
The investment is watching Krutrim-1, which has a 7 billion billion parameter system that serves as the first major language model in January. The placement of the SuperComputer with NVIDIA is planned to broadcast live through the planned expansion in March.
