IndiaAI Mission has recently tapped three new teams — Soket AI, Gnani.ai, and Gan.ai — to develop foundational artificial intelligence models tailored to the country’s linguistic richness and advanced reasoning needs. This announcement, made on Friday, follows the earlier selection of Sarvam.ai for similar foundational AI work.
The fresh selections underline India’s growing push to create homegrown AI technologies that can address the unique challenges and opportunities in its diverse society.
Soket AI to Build a Massive Open-Source Model for Critical Sectors
Soket AI is set to develop a foundational model with a whopping 120 billion parameters, optimized specifically for India’s multitude of languages. This open-source project aims to support vital sectors such as defense, healthcare, and education.
Creating an open-source model at such a scale is no small feat. Soket AI’s approach promises to democratize access to powerful AI tools for Indian developers and researchers. By focusing on multilingual capabilities, they hope to bridge gaps in communication and accessibility that many Indian states face due to language barriers.
The project will likely unlock new AI-driven solutions—imagine health diagnostics in regional dialects or educational content tailored to local languages. Soket AI is carving a path for AI that truly reflects the country’s diversity.
Gnani.ai’s Voice AI Model Brings Real-Time Multilingual Speech Processing
Gnani.ai has been given the green light to build a 14 billion parameter voice AI foundational model. This one is all about real-time multilingual speech processing and advanced reasoning capabilities.
The project focuses on speech, an area where many AI systems stumble due to accents, dialects, or simply the sheer variety of languages spoken. India, with its 22 officially recognized languages and hundreds of dialects, represents a massive challenge here.
Ganesh Gopalan, Gnani.ai’s CEO, highlighted the importance of inclusivity. He said, “We are honoured to be selected under the IndiaAI Mission to develop large language models that truly represent India’s linguistic diversity.” Gopalan emphasized that their voice-to-voice AI models aim to communicate in languages that users actually understand—an essential step for making AI truly transformational in everyday life.
Gan.ai Targets Superhuman Text-to-Speech with a Multilingual Twist
Gan.ai’s task is ambitious—building a 70 billion parameter multilingual foundation model focusing on superhuman text-to-speech (TTS) capabilities. The goal? Push the limits of AI in language processing and communication.
Text-to-speech technology has improved a lot, but often it still sounds robotic or limited to a few dominant languages. Gan.ai wants to change that narrative by crafting TTS that not only sounds natural but can handle the diverse linguistic landscape of India.
By focusing on “superhuman” TTS, Gan.ai could revolutionize everything from accessibility for the visually impaired to voice interfaces in smart homes—especially in Indian languages that have been neglected in tech innovation.
New Initiatives by the IT Ministry Highlight Talent and Research Focus
India’s IT Minister Ashwini Vaishnaw didn’t stop at just announcing new teams. He also unveiled plans for a PhD program and a talent development initiative within the IndiaAI Mission framework.
Why does this matter? Because foundational AI research needs bright minds working at the cutting edge. Encouraging students and researchers to engage deeply with AI ensures India builds a sustainable ecosystem that doesn’t just import tech but innovates itself.
The government’s approach shows they’re thinking long-term—seeding talent and building infrastructure simultaneously.
Expanded GPU Access to Fuel AI Innovation
On the tech resources front, the IndiaAI Mission has opened up access to 34,000 GPUs via the IndiaAI Compute portal. This includes powerful chips from NVIDIA, AMD, AWS, and Intel.
Here’s why that’s a big deal: GPUs are the workhorses behind training complex AI models. Before this, many startups and researchers in India struggled to get enough computational power to train large-scale AI.
Now, with more GPUs available, more innovators can push their projects forward without worrying about crippling costs or delays.
GPU Access Summary Table
| Provider | GPU Quantity | Impact Focus |
|---|---|---|
| NVIDIA | Large pool | AI model training |
| AMD | Moderate | Research and startups |
| AWS | On-demand | Cloud-based AI services |
| Intel | Significant | Diverse AI workloads |
This resource boost will likely accelerate development across all three newly selected teams and others in the IndiaAI ecosystem.
What’s Next for India’s AI Ecosystem?
The IndiaAI Mission’s selection of Soket AI, Gnani.ai, and Gan.ai represents a crucial step in building AI systems that actually speak India’s languages—and think in ways that fit its needs.
It’s exciting to see an initiative that combines the power of big AI models with an acute awareness of cultural and linguistic diversity. In a country as complex as India, AI that ignores this diversity simply won’t work.
The government’s dual approach—combining model-building with talent development and resource expansion—feels like a smart bet for the future.
Whether it’s healthcare in Marathi or defense communication in Tamil, the coming years may bring AI that truly feels Indian, in every sense.
