
Zoho co-founder and chief scientist Sridhar Vembu has voiced his support for AI startup Sarvam AI. The company received online criticism after releasing its new large language model, Sarvam-M. Vembu urged the team to maintain focus and emphasized that no product becomes an immediate hit.
The IndiaAI Mission’s Sarvam AI recently launched a 24B-parameter hybrid model trained on Indic languages, math, and programming. It’s part of its continuous work to create a foundational LLM that is independent. Despite a slow start with downloads for Hugging Face, interest is gradually increasing.
Early Numbers Don’t Define an AI Startup
Mistral Small is the foundation of Sarvam AI’s flagship product, Sarvam-M. This model was first created by the French company Mistral AI. Following that, the AI startup improved it with a focus on Indic languages and reasoning abilities. Despite the technical foundation, Sarvam-M received harsh criticism from investors and developers worldwide. It saw just 334 downloads in its first two days on Hugging Face.
Menlo Ventures partner Deedy Das criticized Sarvam’s performance. Moreover, he contrasted it with a language model developed by two Korean students and made available to the public. Their model received 200,000 downloads in just one month. Sarvam’s progress was deemed “embarrassing” by him, who questioned its direction and potential. There was intense discussion among the Indian tech community about whether this criticism was constructive or dismissive.
Sridhar Vembu Backs the Indian AI Startup Team
Sarvam-M is a hybrid language model with 24 billion parameters. It is designed to be used with Indic text and for more complicated tasks like math and programming. The AI startup highlighted its wider applicability across various domains. Additionally, it highlighted the importance of continuing model development in line with India’s strategic objectives regarding emerging technologies.
In defense, Sridhar Vembu recalled Zoho’s slow entry into the Indian tech market. He said, “No product we built was ever an instant hit.” Furthermore, he stated that it is not appropriate to judge Sarvam’s efforts solely based on initial outcomes. In a post on X (formerly Twitter), he urged people to “keep fighting the good fight.”
Meanwhile, Sarvam co-founder Pratyush Kumar remained optimistic, inviting feedback to improve the model pipeline. “Great to be receiving feedback on Sarvam-M,” he wrote. Additionally, his colleague Aashay Sachdeva responded to criticism by clarifying false information regarding the company’s valuation and stating that Hugging Face metrics frequently lag. At the time, 718 downloads of an image from the Hugging Face page demonstrated the growing interest.
Indian Tech Community Backs Sovereign AI Model
Sarvam AI plays a crucial part in the IndiaAI Mission, which is entrusted with creating India’s sovereign language model. The Indian government’s selection of Sarvam AI is a strong signal of trust in domestic innovation. Therefore, for a new AI startup, this poses both a challenge and an opportunity.
The rise in Hugging Face downloads to over 700 signifies a turning point, despite the initial figures being unimpressive. Researchers, policy experts, and developers are starting to use the model more. As Sarvam evolves into more powerful iterations, this interaction is essential.
Sarvam’s team is working to develop a foundation model that is entirely indigenous in the future. With community involvement and strategic support, the Indian tech startup has the potential to become a key component of the country’s AI capabilities.
What Will Define Sarvam’s Legacy?
Sridhar Vembu wisely observed that instant viral fame rarely signals lasting success in the technology sector. Even though Sarvam AI is off to a slow start, its efforts have the potential to yield significant, long-term results. This AI startup’s journey demonstrates India’s dedication to developing autonomous and cutting-edge technological capabilities. In the future, early modest growth could develop into substantial advancement and influence.