SOKET LABS
Cert-In gets Mythos access, being careful with testing: Officials
Government agencies cannot download the AI’s source code or weights for running on their own, completely isolated or ‘air-gapped’ systems. Instead, to use Mythos, they must send their proprietary data, code, or network blueprints over the internet to Anthropic’s cloud servers.
Soket Labs, ZenteiQ launch indigenous foundation models under IndiaAI Mission
Soket Labs and ZenteiQ introduced new artificial intelligence foundation models under the IndiaAI Mission initiative. These models include EKA-1, Dhir 2B, and Dhir 5B, which were unveiled during a launch event. The event featured key government officials, including MeitY Secretary S Krishnan. This launch is part of the government's efforts to enhance India's indigenous AI capabilities. Such initiatives aim to support the growth of foundational AI technologies within the country.
Voice, vernacular, victory!
Technology companies and startups are betting big on voice AI models. Experts say Indian firms are more likely to crack it with affordable pricing and local context.
Voice AI startups hear the sound of money as enterprises sign up
In January Bolna AI, a Y-Combinator startup, raised $6.3 million, Ringg AI raised 5.5 million and ArrowHead raised $3 million, with Voice AI startups such as Navana AI, Vaani AI and JobsUPI are looking to raise funds, founders told ET.
Voice AI attracts interest; Inside PhonePe’s IPO plan
Happy Thursday! Voice AI startups are beginning the year on a high note. This and more in today’s ETtech Morning Dispatch.
Union Minister Ashwini Vaishnaw discusses manufacturing of sovereign, high-end GPUs in India with Nvidia officials
India is stepping up plans to build its own graphics processors after talks between minister Ashwini Vaishnaw and Nvidia executives. The discussions covered sovereign GPUs and local manufacturing of advanced AI devices. The government is also expanding subsidised access to GPUs to boost domestic AI development.
ET Graphics: AI Summit lays out red carpet for India’s sovereign AI dreams
As India accelerates efforts to build a sovereign AI stack, a dozen startups and institutions backed by the government’s IndiaAI Mission are preparing to launch indigenous foundation models. Many of these launches are timed around the India AI Impact Summit next month.
DPDP rules to be released by 28 September: Vaishnaw
The DPDP rules will be followed by a comprehensive set of frequently asked questions, Vaishnaw said. This will include the ones put to the Ministry of Electronics and Information Technology (MeitY) by 22 press bodies regarding how contentious provisions of the DPDP Act will practically affect journalism in India, other officials said. The DPDP Act had been officially notified more than two years back, following more than a decade of debate, judgements and legislation on how to protect the privacy and data of users while the draft rules were released in January.
Eight local firms may get IndiaAI foundational AI models incentives
The Indian government is set to announce the next round of beneficiaries under the IndiaAI Mission, with close to homegrown AI firms expected to receive incentives. BharatGen, a consortium of IIT researchers, is likely to be among those selected. This initiative aims to boost indigenous AI model development by providing subsidised GPU compute, though GPU availability remains a challenge.
The great Indic data hunt
Race to build Indic models has heated up, but challenge over availability of Indic language data persists.
Government to declare Round 2 beneficiaries of Rs 1,500 crore AI mission next week
Avataar.ai, TechMahindra, Fractal Analytics, Genloop, IntelliHealth, Shodha AI, and Zenteiq Aitech Innovations are likely to be selected. Union IT and electronics minister Ashwini Vaishnaw is expected to announce the beneficiaries selected in the second round of screening next week.
Third tender set to add 3,800 GPUs to IndiaAI GPU cluster
IndiaAI Mission is expanding its GPU resources. Around 3,850 GPUs will be added through the third tender. This includes Google Trillium TPUs and Nvidia H100 GPUs. The total GPU count will reach 34,333. Four companies cleared the technical round. The mission aims to offer GPUs at competitive prices. It supports startups building large language models.
Google’s AI model Gemini 2.5 Flash to support local processing of data
This is part of broader efforts from Google as it continues to invest and grow its footprint in India, Google Cloud Asia Pacific vice-president Bikram Singh Bedi told ET on the sidelines of the Google I/O Connect event here on Wednesday. “A key part of investments is partnerships, the local customers, with the local partners to drive things forward, as far as AI from Google and Google Cloud are concerned,” he said.
Sarvam AI will open source its IndiaAI Mission AI models
The company has received the highest subsidy allocated under the IndiaAI Mission so far at Rs 98.68 crore, out of a bill of Rs 246.71 crore for access to 4,096 Nvidia H100 GPUs for six months, as per the IndiaAI website.
India AI Mission: 43 of 506 foundational AI model proposals target large language models
As part of the mission, the government is incentivising the development of LLMs built by startups such as Sarvam, Gnani, Gan and Soket AI Labs with investment capital and other support. According to people familiar with the matter, the IndiaAI Mission will soon announce more LLM-developing applicants who will be given GPU access or grants.
Sarvam and three other AI firms in MeitY’s LLM build out first shortlist
Soket AI Labs’ founder Abhishek Upperwal said it has proposed to create a 120-billion parameter open-source Indic LLM (large language model) trained on 2 trillion tokens under the ‘EKA Project’. Upperwal did not share the funding requirement. Meanwhile, Gnani.ai and Gan.ai have proposed to build small language models.
ETtech Explainer: Meet AI ‘Saba’ trained on Arabic, Tamil, Malayalam
Indian startups developing Indic AI models face competition from global players like Mistral AI, which launched a multilingual model trained in Arabic, Tamil, and Malayalam. However, regional nuances remain a challenge for these global competitors.
DeepSeek signals Indian startups’ deep AI dive
As the battle between OpenAI and DeepSeek escalates, some Indian start-ups have started integrating DeepSeek with offerings. But the tide turned within a few days as OpenAI released its much-anticipated model, o3-mini, its most cost-efficient reasoning model.
Sops, DeepSeek drive local firms to work on foundation AI models
The government’s subsidised common compute facility and opening for grants and funding under the IndiaAI Mission have also spurred energy among startups seeking capital.
Load More
