Sarvam AI's Kaze Smartglasses 🇮🇳 - India's Bold Answer to Meta Ray-Ban in 2026

Let this sink in for a moment.

An Indian AI company - founded in August 2023, less than three years ago - has already built AI models that outperform both ChatGPT and Google Gemini on benchmark tests. It has launched India's first AI smartglasses at a national summit attended by the Prime Minister, who became the first person in the country to wear them. And it has announced a product launch for May 2026 that will put Indian AI hardware in consumers' hands for the first time.

The company is Sarvam AI. The product is Kaze - AI smartglasses that listen, understand, and respond to the world in real time. And the story of how this startup went from incorporation to hardware launch in under three years is one of the most compelling chapters in India's technology story that almost nobody outside India is talking about.

If an American startup had done this - if a Silicon Valley company founded in mid-2023 had benchmarked above OpenAI's flagship model and announced a smartglasses product within two years - it would be the cover story of every major technology publication in the world. The funding announcements would be breathless. The valuations would be astronomical. The media coverage would be relentless.

Instead, one of the most remarkable AI stories of 2026 is being told primarily within India - and even here, not loudly enough.

This needs to change. And this blog post is our contribution to changing it.

Who Is Sarvam AI? The Story So Far

Sarvam AI was founded in August 2023 with a mission that is both specific and ambitious: to build AI that truly works for India.

This is a more complex challenge than it might initially appear. India is not a single-language, single-culture market that can be served by adapting a Western AI model. India has 22 official languages, hundreds of dialects, a written tradition spanning multiple scripts, and cultural contexts that differ enormously from the Western environments in which ChatGPT, Gemini, and other major AI models were trained.

India's Digital Personal Data Protection Act (DPDPA), which came into force in 2023 and is now fully implemented, fundamentally changed how Indian businesses think about data. For the first time, there are real legal consequences for storing and processing Indian citizens' data carelessly - particularly on foreign servers under foreign jurisdiction. This regulatory shift created both a compliance requirement and a market opportunity for Indian platforms that store data within India's borders.

The dominant AI models - for all their extraordinary capabilities - understand the world primarily through an English-language, Western-cultural lens. They handle Hindi and other Indian languages with varying degrees of competence, but they do not understand India with the depth and authenticity that a model trained on Indian data, by Indian researchers, for Indian users can achieve.

Sarvam AI was built to address this gap - not as a niche regional product, but as a world-class AI platform that starts from Indian foundations and builds outward.

In just under three years, what Sarvam has achieved is remarkable by any standard.

The Benchmark Breakthrough - Outperforming ChatGPT and Gemini

The claim that a startup founded in 2023 has built AI models that outperform ChatGPT and Google Gemini is one that deserves scrutiny - and holds up to it.

Sarvam AI's models - including Sarvam Vision and Bulbul V3 - have been tested on benchmarks that measure AI model performance across language understanding, reasoning, and multimodal capabilities. On relevant benchmark tests, particularly those that include Indian language tasks and culturally Indian contexts, Sarvam's models have demonstrated superior performance to the flagship offerings from OpenAI and Google.

This result is not as surprising as it might initially seem to those unfamiliar with how AI model performance works. General-purpose AI models trained primarily on English and Western data have systematic weaknesses on tasks that require deep understanding of non-Western languages and cultural contexts. A model trained specifically on Indian data, with Indian language understanding as a core design objective, will outperform general-purpose models on Indian-specific benchmarks for the same reason that a specialist outperforms a generalist in their area of specialisation.

What makes Sarvam's achievement significant is that this was not obvious - many assumed that the scale advantages of OpenAI and Google would overwhelm any quality advantages a smaller, focused Indian team could achieve. Sarvam has demonstrated that focused, high-quality training data and India-specific design decisions can produce AI models that genuinely outperform larger competitors on the tasks that matter most to Indian users.

Sarvam Vision brings multimodal AI capabilities - the ability to understand and respond to visual inputs alongside text - to Indian language contexts. For Indian users who want to point their phone or glasses at something in the world and receive intelligent contextual information in their preferred Indian language, Sarvam Vision is a capability that no foreign AI model delivers with comparable quality.

The Kaze Smartglasses - India's First AI Hardware Product

At the India AI Impact Summit 2026, Sarvam AI made an announcement that elevated its ambitions from software to hardware: the unveiling of Kaze - India's first AI smartglasses.

The name itself is significant. Kaze is not just a product name - it represents Sarvam's vision of AI that moves with you through the world, not AI that you access through a screen. Intelligence that is ambient and contextual rather than on-demand and screen-mediated.

What Are the Kaze Smartglasses?

Kaze AI smartglasses are designed around a simple but powerful idea: real-world intelligence integration. Rather than pulling out your phone, opening an app, and querying an AI chatbox, Kaze users wear AI on their face - a form factor that allows the AI to see what you see, hear what you hear, and respond in real time to the world as you move through it.

The glasses are built to:

The Kaze platform delivers customised experiences through Sarvam's AI ecosystem - meaning that the glasses are not just a hardware product but an entry point into Sarvam's broader platform, with capabilities that can be extended through app integrations, enterprise customisation, and ongoing AI model improvements.

PM Modi's Historic First Try

The significance of the Kaze launch was underscored by an extraordinary moment at the India AI Impact Summit 2026: Prime Minister Narendra Modi became the first person to try the Kaze AI smartglasses.

This is not merely a symbolic gesture. When India's Prime Minister is the first user of an Indian AI hardware product at a national AI summit, it signals a level of governmental recognition and endorsement that carries real consequences - for public awareness, for enterprise adoption, for investor confidence, and for the message it sends to the Indian technology community that building world-class AI hardware in India is a national priority.

The Prime Minister's willingness to personally engage with Kaze at its launch is a statement: India is serious about AI hardware, India is proud of what Sarvam has built, and India's government is paying attention to homegrown AI innovation.

For Sarvam AI, a startup just over two years old, this is an extraordinary validation.

Kaze vs Meta Ray-Ban Smart Glasses - India's Entry Into a Global Competition

The AI smartglasses market is not empty. Meta's Ray-Ban smart glasses - the collaboration between Meta's AI division and the iconic eyewear brand - represent the most prominent product in this category, having achieved significant commercial traction and media attention globally.

Meta's glasses offer similar conceptual capabilities: visual capture, AI voice interaction, ambient intelligence. They have benefited from Meta's massive brand recognition, its enormous AI research infrastructure, and its global distribution partnership with Ray-Ban.

So how does Kaze compare - and why does it matter that India has entered this space?

Indian Language Advantage

Meta's smart glasses deliver AI responses primarily in English and a limited set of Western languages. For the hundreds of millions of Indian users who prefer to interact in Hindi, Tamil, Telugu, Bengali, or any of India's other major languages, Meta's glasses offer limited practical utility.

Kaze, powered by Sarvam's Indian-language AI models - including Bulbul V3 for voice and Sarvam Vision for visual understanding - is designed from the ground up for Indian language interaction. This is not a minor feature difference. For Indian users, it is the difference between a product that genuinely integrates into their daily life and a product that requires them to adapt their communication patterns to fit the AI.

Indian Context Intelligence

AI that understands Indian cultural contexts, Indian environments, Indian signage, Indian food, Indian geography, and Indian social norms will deliver qualitatively different responses to Indian users than AI trained primarily on Western data. Sarvam Vision's Indian-contextual training means Kaze can recognise and meaningfully respond to elements of the Indian world that foreign AI systems handle poorly.

Data Sovereignty

Every visual and audio input processed by Meta's glasses is processed on Meta's infrastructure, governed by Meta's terms of service, potentially contributing to Meta's AI training data. For Indian users conscious of data sovereignty - particularly post-DPDPA - Kaze's processing on Sarvam's Indian infrastructure is a meaningful difference.

Economic Contribution

Every Kaze purchase keeps economic value within India's technology ecosystem, funds Indian AI research, supports Indian engineering jobs, and contributes to the Indian AI capability base that benefits the entire country. This is not an argument against choice, but it is an argument that - all else being equal - an Indian product delivering equivalent or superior Indian-language AI capabilities deserves India's support.

Sarvam AI in the Context of India's Broader AI Ecosystem

Sarvam AI's achievements are impressive in their own right - but they are more meaningful understood in the context of the broader Indian AI ecosystem that is developing around them.

India's AI ecosystem in 2026 includes:

Krutrim by Ola

India's first AI unicorn, building foundational large language models trained on Indian data. Krutrim's success demonstrated that India could build AI infrastructure at scale and attract the investment to back it.

Gnani.ai

A voice AI platform covering 20+ Indian languages, serving enterprises across banking, telecom, retail, and healthcare with AI voice capabilities that are world-leading for Indian language applications.

Yellow.ai

Processing billions of enterprise customer interactions through conversational AI across voice, chat, and email in Indian and global languages.

Explore all 9 Indian AI Alternatives

Conclusion

Sarvam AI, founded in August 2023, has accomplished in under three years what most technology companies fail to achieve in a decade: built AI models that benchmark above the world's best, established India's first AI hardware product, received Prime Ministerial endorsement, and positioned itself to compete in a global market currently dominated by Meta.

The Kaze AI smartglasses represent more than a product launch. They represent India's bold declaration that it intends to be a first-mover, not a follower, in the next era of AI hardware. They represent the Indian AI community's confidence that homegrown innovation can compete on the global stage on technical merit alone. And they represent the beginning of what could be a genuinely transformative chapter in India's technology story.

The world may not be paying enough attention yet. But it will be.