China’s viral AI model DeekSeek appears to have finally spurred Indian AI companies into action.
Ola Krutrim, which is India’s only AI unicorn, has made DeepSeek’s models live on its own cloud platform. Ola claims that it is making the Chinese model available at one of the lowest prices in the world. Krutrim has also said that it will release the details of its own state-of-the-art model on 4th February.
“India can’t be left behind in AI,” Ola Krutrim founder Bhavish Agarwal posted on X. “Krutrim has accelerated efforts to develop world class AI. As first step, our cloud now has DeepSeek models live, hosted on Indian servers. Pricing lowest in the world. Details on our AI lab, SOTA model and research progress, open source drops on 4th Feb!” he added.
Krutrim has made 5 versions of DeepsSeek’s models available on its site. These include the DeepSeek-R1-Distill-Llama-70B, DeepSeek-R1-Distill-Llama-8B, DeepSeek-R1-Distill-Qwen-14B, DeepSeek-R1-Distill-Qwen-32B and the DeepSeek R1. The prices for these models range from Rs. 40 to Rs. 60 to generate 1 million tokens. China’s DeepSeek is itself offering its model at $0.55 (Rs. 40) per million input tokens and $2.19 (Rs. 180) per 1 million output tokens from its own servers.
Krutrim isn’t the first AI company to offer DeepSeek’s models on its platform. DeepSeek’s model is open-source, and other companies too have begun offering its outputs. Perplexity, in particular, was the first major company to offer DeepSeek on its platform. Founder and CEO Aravind Srinavas had said that the version of Perplexity it was offering ensured that user data stayed within US servers, and the model was also not hampered by many of the guardrails of the original Chinese model, which refused to answer questions on the Tianaamen Square massacre or whether Arunachal Pradesh was a part of India.
Ola Krutrim’s more interesting announcement is about its own model. Bhavish Agarwal hinted that the model would be state-of-the-art, which indicated that it would compete with the best global models in its class. He also said it would be open-source. This would be an interesting choice — most top US companies like OpenAI and Anthropic have chosen to not make their model weights public, but DeepSeek had released its models as open-source with an MIT license. After DeepSeek was released, it garnered the attention of governments everywhere — US President Donald Trump said that it was a wake-up call for the US tech industry, and India’s IT Minister had praised the Chinese model and said it was “impressive”. This had led to Indians call for the development of their own models, and with Krutrim set to announce its own model in 4 days, India might be finally looking to take its first steps in the AI race.