Co-Founder Sarvam AI
Bengaluru
Joined August 2010
- Tweets32
- Following40
- Followers4.5K
- Likes30
Vivek Raghavan retweeted
As part of efforts towards infusion of Artificial Intelligence in training, a first of its kind customised GPT named #DRONA (DSSC Resource Optimising Neural Al) ver 1.0 was indigenously developed by Defence Service Staff College Wellington and was launched on #DSSC WAN on 16 Jun 2025.
Towards keeping pace with advancements in the domain of Generative Al and to incorporate indigenous technological stacks, a multi modal DRONA 2.0 GPT was developed integrating the indigenous SARVAM 105b LLM along with open source LLMs. #DRONA 2.0 was launched today by Lt Gen Manish Erry, Commandant #DSSC in the presence of Dr Vivek Raghavan, Co-Founder, Sarvam Al. Along with the launch, a Memorandum of Understanding was signed between #DSSC and SARVAM Al for collaboration and further development of DRONA GPT & various Al based applications.
#DRONA 2.0 is hosted on state of the art GPU server for use in DSSC & has multimodal features like reasoning, document inferencing, image recognition/ generation, text to speech/ speech to text, translation/ transliteration, podcast/ ppt generation and tools for data analytics. Being the first of its kind multi modal GPT in armed forces, the application was subjected to extensive audit and was certified by #HQIDS/ Defence Cyber Agency (DCyA).
DRONA 2.0 is pre-trained on a large repository of the academic content residing at DSSC providing major fillip to faculty, students of DSSC, aiding them towards access of information and progress advanced studies/research in the field of Defence and Strategic Studies.
#artificialintellingence
#DRONA2.0
#professionalmilitaryeducation
@DefenceMinIndia
@SethSanjayMP
@SpokespersonMoD
@MIB_India
Stay Tuned …
Sarvam Epoch | Chapter 1
Join us for the first edition of Sarvam Epoch, our flagship AI conference, in Bengaluru.
Across the two days, we will share new model updates, launch new products, present our latest research, and host hands-on technical workshops.
Builder Edition, 30 July | Enterprise Edition, 31 July
Register here: epoch.sarvam.ai/
Vivek Raghavan retweeted
📢 Open-sourcing the Sarvam 30B and 105B models! Trained from scratch with all data, model research and inference optimisation done in-house, these models punch above their weight in most global benchmarks plus excel in Indian languages.
Get the weights at Hugging Face and AIKosh. Thanks to the good folks at SGLang for day 0 support, vLLM support coming soon. Links, benchmark scores, examples, and more in our blog - sarvam.ai/blogs/sarvam-30b-1…
Vivek Raghavan retweeted
Drop 14a/14: Over the past few days, we’ve been putting our models, products, and research out there. The positive feedback has helped more people agree with our long held belief that #IndiaCan be a builder in this space.
Today, we are rolling out Indus - a chat interface to experience Sarvam 105B. Here is what all you can do:
Vivek Raghavan retweeted
Sarvam AI provides the momentum needed to compete globally while building for Indian diversity.
It advances India’s full-stack AI journey from foundational models to real-world deployments. The 105B and 30B parameter model launches signal growing depth in India’s AI ecosystem.
Wish the team continued success.
@pratykumar @vivekrag @aalekhsharan @sarvamai
Vivek Raghavan retweeted
Drop 10/14: Announcing Sarvam Edge, our dedicated effort to bring intelligence to run offline and on-device. Our goal is to make AI that is efficient, private, and accessible everywhere.
Today we are showcasing our speech recognition, speech synthesis, translation, and document digitisation models running on-device. This comes on the back of our research efforts to make models super small in memory and compute footprints, while being close to accuracy of much larger models.
We believe running models at the edge will truly unlock population scale AI in India. And we are working with partners to make this a reality. Read more in our blog here: sarvam.ai/blogs/sarvam-edge
Vivek Raghavan retweeted
Drop 8/14: Introducing Saaras V3, the next iteration of our speech recognition model. We extend our lead in this space with an even more accurate model, particularly for mixed-language and noisy speech.
We have also expanded support for all the 22 scheduled languages of India.
And we now support real-time streaming! In this mode, the model delivers low latency while preserving transcription quality.
sarvam.ai/blogs/asr
Vivek Raghavan retweeted
The all-new Sarvam is here.
Designed for all of India.
sarvam.ai/brand
Vivek Raghavan retweeted
Drop 5/14: Introducing Bulbul V3, our latest text-to-speech model. It raises the bar for how human it sounds, while being super robust.
In an independent third-party human listening study, Bulbul V3 delivers the highest listener preference, and low error rates across use-cases and languages. See details in our blog, but first watch the video. sarvam.ai/blogs/bulbul-v3
Vivek Raghavan retweeted
Special shoutout to the incredible women who have shaped Samvaad.
Especially @meenu_stardust, who’s led engineering since day one. She's usually hustling quietly behind the scenes but Samvaad wouldn’t be what it is without her.💛
Samvaad is where it all started for me, that's why this announcement is so special! Seeing us scale from those early days to powering real businesses today feels incredibly full-circle. So proud of this team @SarvamAI and this journey 🫶
Vivek Raghavan retweeted
Drop 4/14: Introducing Sarvam Vision: a state-space based 3 billion parameter vision language model that is competitive with the best results in digitisation in English, and defines a significantly higher bar for Indian languages. See the details in our blog: sarvam.ai/blogs/Sarvam-visio…
Vivek Raghavan retweeted
Drop 3/14: Our Conversational Agents platform, Samvaad, now powers over a million minutes of interactions daily. Every month our TAM estimates keep increasing as we discover new use cases, not thought possible earlier - population scale outreaches, hybrid onboarding journeys over phone and WhatsApp, 24*7 sales assistants, ...
Here is a real conversation between a merchant and B2B marketplace support agent.
Vivek Raghavan retweeted
Sarvam Audio works great for multi-speaker transcription with high quality separation between speakers achieving lowest diarisation error rates across languages.
Vivek Raghavan retweeted
Sarvam Audio is trained on top of Sarvam 3B a large language model trained from scratch by @SarvamAI supporting 22 Indian languages and English. The model aces fine control of transcripts amongst various different formats, achieving the lowest word error rates across languages.
Vivek Raghavan retweeted
Drop 2/14: Sarvam Audio: a state-space based efficient audio language model that defines the new benchmarks in speech recognition for Indian languages. Significantly outperforms Gemini 3 and GPT 4o Transcribe in a range of benchmarks. See details in our blog: sarvam.ai/blogs/sarvam-audio
@SarvamAI
Vivek Raghavan retweeted
Drop 1/14: We pulled off a first of its kind. The budget speech by the Finance Minister was live telecast in multiple languages on @republic. Reached millions of homes with a latency under 2 minutes. youtube.com/watch?v=wiQ9KesT…
We did more than 7.5 million calls/WhatsApp conversations on Samvaad last month.
Introducing Sarvam Samvaad 🚀
Sarvam Samvaad, our Conversational AI Platform, is designed to help enterprises build, test, and launch AI agents fluent in 11 Indian languages.
▪️Power interactions across telephone, WhatsApp, web, and apps
▪️Handle complex phrases, alphanumerics, and proper nouns with precision
▪️Listen in on every conversation and discover deep insight
▪️Scale your support with pricing built for India
Trusted by leading brands across the country, the platform delivers production-ready agents so enterprises can move from pilot to full-scale deployment in just days.
It’s time to level up customer experience with Sarvam Samvaad.
Watch it in action ↓
Our newest contribution to India’s AI ecosystem. Pls check it out.
Today we’re announcing Sarvam-Translate, an open-weights model that translates text across 22 Indian languages, with support for long-form text and the ability to handle diverse formats, contexts, and styles.
Sarvam-Translate stands out for its ability to handle the complexities of real-world content. The model excels at translating web pages while preserving structure, interpreting idioms and slang, translating social media posts with emojis, and preserving syntax when translating code files.
Sarvam-Translate achieves significant improvements over much larger models on key Indian language translation benchmarks.
▪️️In human evaluations conducted by language experts, Sarvam-Translate consistently outperformed much larger models, including Gemma3-27B-IT, Llama4 Scout, and Llama-3.1-405B
▪️In automated evaluation on ability to conform to structured long-form content, Sarvam-Translate shows high accuracy (> 4.9 over 5) for 15 languages
Here’s a detailed blog showcasing how Sarvam-Translate performs in real-world translation scenarios, with examples in action and benchmark results - sarvam.ai/blogs/sarvam-trans…
You can now try Sarvam-Translate on the Sarvam API - dashboard.sarvam.ai/translat…
Watch this space for more updates!
Sarvam-M represents an important stepping stone on our journey to build Sovereign AI for India. More to come ….
Today we introduce Sarvam-M, a 24B open-weights hybrid model built on top of Mistral Small.
Sarvam-M achieves a new benchmark across a range of Indian languages, math, and programming tasks, for a model of its size.
Here is a detailed technical blog on how we customize fine-tuning and reinforcement learning recipes, and optimize deployment - sarvam.ai/blogs/sarvam-m
Sarvam-M is built for versatility and designed to support a wide range of applications, including conversational agents, translation, educational tools, and so much more.
You can now try Sarvam-M on the Sarvam API - dashboard.sarvam.ai/playgrou…
This is the first in a series of contributions as we help build out the Sovereign AI Ecosystem in India.
Stay tuned for more updates in the coming weeks.
