India’s efforts to build its large language model (LLM) got a significant push this month with the release of Sarvam-M.
It is a 24-billion parameter open-weights AI model from Bengaluru-based startup Sarvam AI, one of the key players selected under the IndiaAI Mission.
But the release also sparked a debate: is this truly a breakthrough or just repackaged hype? Here's what you need to know.
Sarvam-M (M stands for Mistral) is a “hybrid” language model, meaning it’s trained on a blend of tasks such as math, programming, reasoning and Indian language understanding.
It is built on top of the open-source Mistral Small model and fine-tuned for Indian use.
The model is designed to power applications such as chatbots, translation systems and educational tools and is available through Sarvam’s API and on Hugging Face.
"Sarvam-M represents an important stepping stone on our journey to build Sovereign AI for India," co-founder Vivek Raghavan said on X, formerly Twitter.
Sarvam-M's Training Process: (SFT), (RLVR) & Inference Optimisation
Let's Understand one by one:
1. Supervised Fine-Tuning (SFT): Trained on high-quality Q&A pairs to improve relevance, reduce bias, and ensure cultural appropriateness, enhancing both conversational and problem-solving abilities.
2. Reinforcement Learning (RLVR): Refined using instruction, programming, and math data via feedback loops, boosting logical reasoning and instruction-following skills.
What can Sarvam-M do?
– Conversational AI: Powers chatbots and virtual assistants. – Machine Translation: Supports English ↔ Indian language translation.
What can Sarvam-M do?
– Education: Solves math/science problems, aids competitive exam prep (e.g., JEE Advanced questions in Hindi via "Think" mode), enhancing accessibility for Indian students.
So, why the backlash?
Despite all the technical achievements, the model didn’t get the warm welcome one might expect.
On Hugging Face, a platform where developers can download and test AI models, Sarvam-M was downloaded only 334 times in the first two days. Some critics saw this as a sign of failure.
This sparked a debate. Supporters of Sarvam-M, including Aashay Sachdeva from the company, defended the model, highlighting its benchmark results and customisation process.
Meanwhile, Sarvam’s co-founder, Vivek Raghavan, called Sarvam-M a “stepping stone” toward building India’s own AI systems.
The company is one of the few chosen under the Indian government’s IndiaAI Mission to develop a sovereign LLM for the country.
Zoho founder Sridhar Vembu also urged people not to focus only on instant success.
He said that most products take time to find their place, and praised Sarvam for their efforts. “Keep fighting the good fight,” he encouraged.