• Krutrim, an Indian AI unicorn led by Bhavish Aggarwal, has become the first Indian AI company to host Meta's Llama 4 models on India-based servers.
  • The company is expanding its AI offerings with plans to release Krutrim Assistant v2 with "DeepSearch" capability this month, while also aiming to scale its data center capacity to 1 GW by 2028.

Bhavish Aggarwal-led AI unicorn Krutrim has announced that it has started hosting Meta’s Llama 4 models—Scout and Maverick—on its cloud infrastructure in India.

With this, the startup claims to be the first Indian AI company to deploy Meta’s Llama 4 models on servers hosted within the country.

Llama 4 Scout is a 17 billion active parameter model featuring 16 experts and a 10 million token context, while Llama 4 Maverick also has 17 billion active parameters but with 128 experts and a 1 million token context window.

Developers can access these models on Krutrim’s platform at a price ranging from INR 7 to INR 17 per million tokens.

“Excited to share that @Krutrim is among the world’s first to host Meta’s Llama 4 models running entirely on it’s India-hosted cloud. Powering our developers with world-class AI, at industry-disrupting prices, with complete data sovereignty,” Aggarwal shared on X.

This move adds to Krutrim’s expanding offerings, which include GPU-as-a-service, model-as-a-service, and a suite of no-code AI tools.

The startup had earlier deployed open-source AI models by the Chinese GenAI company DeepSeek, including the R1 671B model on Nvidia H100 GPUs.

The startup reiterated that the second version of its Krutrim Assistant is scheduled for release later this month. This new version will come equipped with “DeepSearch,” a tool developed to enhance the accuracy and efficiency of data search.

Looking ahead, Krutrim has set ambitious goals to scale its data centre capacity to 1 GW by 2028. It has also submitted a proposal under the INR 10,037 Cr IndiaAI Mission to build indigenous AI foundational models. The government is expected to announce selected proposals shortly.

Edited by Annette George