Indian AI startup launches Sarvam-M model: What is it, why is everyone talking about it
India's homegrown AI startup Sarvam has launched its newest language model, Sarvam-M, which is making waves in the tech community for both good and not-so-good reasons. The model is being praised for its focus on Indian languages, maths, and programming tasks, but it is also facing criticism for not being 'good enough.' The drama surrounding the AI company has sparked even more interest. If you have questions too, here's a breakdown of what the Sarvam-M model is, why it matters, and why the AI company is facing backlash.advertisementWhat exactly is Sarvam-M?Sarvam-M is a large language model, or LLM, developed by Indian startup Sarvam AI. These types of models are trained to understand and generate human-like text, and they power tools like chatbots, translation software, and educational apps. Sarvam-M is based on a smaller model called Mistral Small and has been expanded into a much larger system with 24 billion parameters, which are basically the knobs and dials that help it process language and learn from data.In simple terms, Sarvam-M is like a very smart AI assistant that can handle a wide range of tasks, from answering complex math questions to understanding and responding in Indian languages like Hindi, Bengali, Gujarati, and more. What makes it different is that it has been built with India in mind, supporting 10 local languages and offering strong performance in tasks involving both language and reasoning.How was it built?advertisement
Sarvam-M was trained using a three-step process:Supervised Fine-Tuning (SFT): This stage involved feeding the model high-quality questions and answers to help it learn. The team made sure the responses were relevant, less biased, and culturally appropriate. This helped the model get good at both everyday conversations and more complex problem-solving tasks.Reinforcement Learning with Verifiable Rewards (RLVR): In this step, Sarvam-M was further improved using data related to instructions, programming, and mathematics. It was taught to follow instructions better and think more logically using feedback loops and carefully designed tasks.Inference Optimisation: This final stage involved making the model run faster and more efficiently. Techniques like FP8 quantisation (a way of simplifying data without losing accuracy) and better decoding methods helped improve the model's speed and performance, though there were still some issues with handling high traffic.What can Sarvam-M do?The model has been built to power various real-world applications. It can be used for:Conversational AI, which will basically mean that it can power chatbots and virtual assistantsMachine translation, where it could be used to translate between English and Indian languagesEducation, considering its ability to solve maths problems, answer science questions, or even helping students prepare for competitive exams like JEE. In fact, one of the Sarvam team members shared results showing that Sarvam-M's "Think" mode correctly answered several JEE Advanced-level questions in Hindi, which can be a major step in making such tools useful for Indian students.How does it compare with other models?advertisementSarvam-M has shown impressive results in certain areas. In a test that combined math with romanised Indian languages, the model achieved over 86 per cent improvement, beating out some other well-known models. It performed better than Meta's Llama-4 Scout on many benchmarks and was on par with much larger models like Llama-3.3 70B and Google's Gemma 3 27B.However, it did slightly underperform in English knowledge tests, with about 1 per cent lower accuracy compared to others. Still, the model stands out for its Indian language skills and reasoning abilities.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.Deedy Das, an investor at Menlo Ventures, called the response 'embarrassing,' saying there's little interest in this kind of work. He compared it to a different model created by two Korean college students, which got nearly 200,000 downloads quickly.
advertisementThis sparked a debate. Supporters of Sarvam-M, including Aashay Sachdeva from the company, defended the model, highlighting its benchmark results and customisation process. He even posted proof of the model's performance on social media.
Another user, who works at AI4Bharat, added that the real achievement was not just the model, but the method used to train it. He said it sets a strong foundation for other Indian developers to build on.
advertisementMeanwhile, 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.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Time of India
an hour ago
- Time of India
‘Veena' guides candidates through UG form fill-up, subject selection
Kolkata: Veena, the new chatbot on the centralised admission portal for undergraduate admissions had interacted with more than 7,236 users and answered 19,844 questions over four days till Saturday evening, showed higher education department data. From helping applicants with selecting colleges and courses to answering queries on the admission process, document verification and payments, Veena, with multilingual support system, seemed to have made it easier for applicants to fill up forms. TOI reported on May 15 on the chatbot's debut on the centralised portal, launched to guide candidates through the process and clear their doubts in real-time, which in turn would ensure fewer errors in the forms. "As the chatbot utilises Domain-Specific Natural Language Processing techniques to understand and process queries in natural language, students can interact in a conversation-like manner in multiple languages, including Bengali, Hindi and English. Some applicants asked questions in Bengali though they wrote it in English font. The chatbot gave the answer in Bengali font," said an official. "The chatbot maintained 35%-38% Bengali support on all days." Another official pointed out that 2.8 interactions with the chatbot per user took place on an average, indicating "strong engagement". "User growth increased in the first few days, and the applicants rated their experience on the chatbot highly, marking a score between 4.5 and 5, 5 being the maximum," said the official. The chatbot was introduced after many of last year's applicants had said they found it difficult to understand the course combinations and many even made some basic mistakes. "But the chatbot helped clear doubts and minimise errors. The queries are changing from quantitative questions, like the number of seats, eligibility criteria and documents required, to qualitative ones, like comparisons between courses and colleges or and best-suited subject combinations. The chatbot is answering the queries precisely and instantly," said an official. The higher education department said the first few days' chatbot interactions showed remarkable AI system evolution as the precision of answers improved, reaching almost 99% accuracy. The interaction data and feedback showed that the AI-driven tool had the capacity for self-learning, improving answers and enhancing the admission process, said an official. "As the chatbot handles a large number of users at the same time, the call waiting-time has minimised. It is ensuring reliability while complying with privacy and security," he said.
Indian Express
an hour ago
- Indian Express
Nalanda University adds 4 new PG courses, V-C says reinforcing its role as centre of high learning
Nalanda University Vice-Chancellor Sachin Chaturvedi has announced four new postgraduate courses in economics, hindi, mathematics and philosophy, taking the total PG programmes to 12. Chaturvedi, who is also the Director General of Research and Information System for Developing Countries (RIS), joined as the university's V-C last month. He said, 'Expansion of courses are meant to expand academic scope and reinforce NU's role as a premier centre of research and high learning.' The new courses are in consonance with the blended vision of ancient and new Nalanda universities to focus on mathematics and philosophy. Meanwhile, Hindi has been added as part of the Central government's thrust to promote mother tongue, and economics is added as per the V-C's vision. 'The spirit of Nalanda has evolved as a vibrant expression of continuous engagement with eternal values and the civilisational ethos of the east. It embodies a rich intellectual tradition, marked by openness, inclusivity and a globally oriented outlook. Building upon this legacy, the vision for the new Nalanda is shaped by a forward-looking blueprint rooted in integrative wisdom. It draws inspiration from an eastern worldview in which faith, intellectual inquiry and material well-being are not regarded in isolation, but as deeply interwoven dimensions of holistic human existence.' The previous PG courses included historical studies, ecology and environment studies, Buddhist studies, comparative religions, Hindu studies (Sanatana Dharma), world literature, archaeology, international relations and peace studies, and MBA in sustainable development and management. NU, established under Nalanda University Act, 2010, started functioning from 2014 with two PG courses having only 12 students. PM Narendra Modi formally inaugurated NU campus — spread over 455 acres at Rajgir foothills — last year. Santosh Singh is a Senior Assistant Editor with The Indian Express since June 2008. He covers Bihar with main focus on politics, society and governance. Investigative and explanatory stories are also his forte. Singh has 25 years of experience in print journalism covering Bihar, Delhi, Madhya Pradesh and Karnataka. ... Read More
United News of India
2 hours ago
- United News of India
Afghan dry fruit traders concerned over continued closure of Attari-Wagah border
Kabul, June 21 (UNI) The closure of the Attari–Wagah border crossing, one of Afghanistan's key trade routes to India, has raised serious concerns among dried fruit exporters of Afghanistan. The Kabul Dry Fruit Exporters Union has warned that stopping freight truck movement on this route could severely disrupt both fresh and dried fruit exports, Tolo News reported. The union has urged Pakistani and Indian authorities to reopen the route permanently for Afghan cargo trucks. Khalid Rahmani, spokesperson for the union, told TOLOnews, 'We hope alternative links through Wagah and Chabahar ports will open soon. Given the current political and trade tensions between Pakistan and India, we predict a decline in exports to both countries.' After the Indian and Pakistani military tensions eased, the border reopened briefly on May 9, allowing about 162 trucks to pass, but it was shut again. At least 12 Afghan export trucks remain stuck at the crossing. With the fruit harvest season approaching, traders warn that continued closure could pose major challenges for Afghan exports. They are calling for diplomatic engagement to reopen this critical trade route. Omid Haidari, head of international relations at the Chamber of Agriculture and Livestock, said: 'We urge the leadership of the Islamic Emirate, especially the Ministry of Foreign Affairs, to strengthen contact with the Indian Foreign Ministry so that our major trade route can be reopened.' Merchant Mahboobullah Mohammadi, said, 'Our biggest export is dried fruit, especially pine nuts, which we mostly send to India via Wagah. But the decline in exports this year has brought significant losses. Our air corridors are still shut, and large amounts of pine nuts remain unsold in local markets.' The Wagah border was closed after tensions escalated between India and Pakistan following the terrorist attack in Pahalgam, India, on April 22. The closure of the land border check post continues to cast a shadow over Afghanistan's regional trade. UNI ANV RN
