Sarvam AI Unveils Trillion-Parameter Model Plans, Challenges ChatGPT and Gemini at 5x Lower Cost

Sarvam AI plans trillion-parameter model, sets sights on competing with ChatGPT, Gemini and Claude

Sarvam AI plans a trillion-parameter model to rival global AI leaders. The company's existing voice AI is cost-effective and performs well. An office in the US will recruit Indian-origin AI researchers. Sarvam AI aims to manufacture intelligence and achieve technological self-reliance. New products and platforms were unveiled at their developer conference. AI startup Sarvam AI has announced plans to build a trillion-parameter artificial intelligence model as it looks to compete with global frontier AI models such as OpenAI's ChatGPT, Google's Gemini and Anthropic's Claude, according to a report by The Times of India. Speaking at the company's developer conference in Bengaluru on Thursday, Sarvam AI co-founder Pratyush Kumar said the proposed model is being designed for advanced use cases including coding, cybersecurity, scientific research and simulations. According to the TOI report, the company's existing 100-billion-parameter model is currently focused on voice AI, conversational systems, English speech recognition, dictation and simulation-based tasks. Kumar said the platform has processed over 325 million minutes of voice calls and, in one benchmark, performed within four to five percentage points of a significantly larger AI model. Also read: Art of AI war: Sarvam to boost defence prowess Kumar also said Sarvam's voice AI systems are four to five times cheaper than comparable global offerings, including for English speech recognition. As part of its expansion strategy, the company is establishing a US office to recruit Indian-origin AI researchers working overseas. "Our goal is to ensure India does not just consume AI but also manufactures intelligence," Kumar was quoted as saying by The Times of India. He added that relying on overseas AI providers could result in dependence not only in terms of costs but also data. Co-founder Vivek Raghavan said the company aims to build what he described as a "real token factory in India", arguing that serving AI inference from within the country is an important step towards technological self-reliance, according to the report. Kumar also said Indian AI products should compete globally rather than seek acceptance solely because they are domestically developed. "Sovereignty should not be a tax," he said. As part of its research push, Sarvam AI has appointed AI researcher Devendra Chaplot, a founding member of Mistral AI and Thinking Machines Lab, as an advisor. Chaplot is based in California. The company also unveiled several new products and platforms at the event, including an India-hosted inference platform, AI-powered smart glasses under its wearables brand Kaze, a Python SDK for model training, the Bulbul V4 multilingual text-to-speech model, Sarvam Vision 2.0 vision-language model, and Indus, an agentic AI platform integrating work, voice and coding capabilities. According to the TOI report, Sarvam AI's strongest commercial traction currently comes from the banking and financial services (BFSI) sector. Kumar said financial institutions are deploying the company's AI stack for voice-based and agent-driven workflows, including an AI agent used by SBI Life's sales network. The company is also seeing demand from digital-native businesses, defence and intelligence organisations, and the public sector.

Everything Sarvam AI announced at its Epoch Builder Edition

Sarvam AI launched new models and infrastructure tools at its Epoch Builder Edition event. The company plans to develop a trillion-parameter artificial intelligence model. New offerings include a text-to-speech system and a coding model. A developer platform will support AI applications in over ten Indian languages. Sarvam AI also announced plans for Sarvam Inference and a San Francisco office. Indian AI startup Sarvam AI on Thursday unveiled a series of new models and infrastructure tools at its Epoch Builder Edition event in Bengaluru, including plans to develop a trillion-parameter artificial intelligence model, a new text-to-speech system, a coding model, and platforms for developers building AI applications in Indian languages.Trillion-parameter modelThe Indian AI startup said it is working on a trillion-parameter model, placing

Sarvam Eyes 1 trillion parameter AI model, To Price It 5.5 Times Cheaper than Global Rivals

The company says the new model will be available within six months and revealed that they were building them from scratch In what will be a boost to India's sovereignty dreams in artificial intelligence, Bengaluru-based startup Sarvam.AI announced a major expansion of its tech stack, introduced new models and developer tools, and brought in new enterprise products. All this while promising to offer the solutions at costs that is 5.5 times lower than global rivals. Speaking at the company's inaugural Epoch conference in Bengaluru earlier today, the company, which saw a $300 million Series-B round led by HCLTech after President Donald Trump's move to ban export of latest AI models from Anthropic and OpenAI, the startup reaffirmed that its one-trillion-parameter model should arrive within six months. In fact, it was barely a couple of days after Sarvam AI CEO Pratyush Kumar had called out the White House for blocking access to Claude Mythos and Fable5 AI models. Their latest fundraise led by HCLTech and participated in by Bessemer Venture Partners, Khosla Ventures and Peak XV Partners takes the company's valuation to $1.5 billion. Kumar's had said he felt countries now cannot afford to mistake access to cutting edge AI systems for true technological ownership. "Fable 5 ban is a good instigation for more people to engage in recognising the need for sovereignty," he had said in response to Zoho founder Sridhar Vembu who declared that "globalisation was dead" post the US export ban. In fact, Zoho had then launched Nathu La, its first in-house server platform with a focus on reducing AI infrastructure costs via datacentre indigenisation. Earlier at the conference today, Pratyush Kumar said "We are very happy to announce that we are building a trillion-plus parameter model right here in India. We are building them from scratch to be competitive in coding, cybersecurity, simulation, science and more." Parameters refer to the dataset used to train AI. The more the parameters the stronger the AI can perform. For context, Kimi K3, the largest open-weight model right now, is a 2.8 trillion parameter model. Put in context, once Sarvam releases its trillion parameter model, they would find themselves bridging the gap with leading AI labs like Anthropic and OpenAI.

Sarvam readying end-to-end AI stack to challenge global giants

The Full-Stack Blueprint Homegrown AI challenger's founders say infra and software layer on models are key to move towards the agentic ecosystem Homegrown AI firm Sarvam has put forward a strong end-to-end AI stack strategy spanning foundational models, software services with cyber and defence capabilities, and even hardware like smart glasses as it looks to compete with giants such as Anthropic, Google and OpenAI.Its large language models will be armed with Indian language capabilities and priced five to ten times cheaper than models like OpenAI's ChatGPT or Google's Gemini, Sarvam

Indian Startup Sarvam AI Bets Big on 1T AI Model

Sarvam AI is building a 1-trillion-parameter AI model and has hired a former xAI researcher as it aims to compete with some of the world's biggest AI companies. Indian AI startup Sarvam AI is making a bold move in the AI industry. At its Epoch 2026 event, Bengaluru-based startup revealed that it is building a one-trillion-parameter foundation model from scratch. This platform will allow developers and enterprises to build AI models tailored for Indian languages and use cases. During the event, co-founder Pratyush Kumar said the model the company is currently working on will target coding, cybersecurity, scientific research, and simulation workloads. Additionally, the company revealed the price for its Sarvam 105B model. As per the announcement, the model will cost $0.80 per one million blended tokens, which is cheaper than global LLMs. In addition, the company has announced partnerships with three IITs and two state governments for pilot projects in education and public service delivery. In the conclusion note, the startup has announced that an AI researcher, Devendra Singh Chaplot, recently joined the team as a part-time advisor, who previously was part of . He was part of founding teams at Mistral AI and Mira Murati's Thinking Machines Lab.

Sarvam is building 1T AI model, hires Elon Musk's xAI expert to take on OpenAI and Anthropic

It also revealed lower pricing and added a well-known AI researcher to its team. Sarvam AI has announced its biggest expansion yet as the Bengaluru-based startup looks to strengthen India's position in the global artificial intelligence race. At its Epoch 2026 event, the company said it is building a one-trillion-parameter foundation model from scratch in India while unveiling a new platform that will allow developers and enterprises to build AI models tailored for Indian languages and use cases. The startup also announced aggressive pricing for its existing AI models, claiming they are significantly cheaper than competing global offerings. Adding to the announcements, Sarvam brought onboard AI researcher Devendra Singh Chaplot, who most recently worked at Elon Musk's xAI, marking a major talent gain for the company. Sarvam built one trillion parameter model Speaking at the launch event in Bengaluru, co-founder Pratyush Kumar said Sarvam AI is developing a one-trillion-parameter foundation model in India that will target coding, cybersecurity, scientific research and simulation workloads. He further added that the model is being built from scratch and will be competing with leading global AI systems. The company also announced pricing for its Sarvam 105B model, stating that the model costs $0.80 per one million blended tokens, compared with $4.50 for OpenAI's GPT-5.4 Mini and $9 for Gemini 3.5 Flash. According to Sarvam, this makes its offering about 5.5 times cheaper than GPT-5.4 Mini and more than 11 times cheaper than Gemini 3.5 Flash. The company launched Sarvam Vision Edge, a document intelligence platform, alongside Sarvam Vision 2.0, an improved optical character recognition for enterprise document processing, during the event. Talking about the new tools, Kumar said that the Odisha government is already using the technology to digitise and extract information from land records. Also read: Asus Pad T3201 with 12.2-inch OLED display launched in India: Price, specifications and availability Sarvam launches platform to help build India-centric AI models Aside from the abovementioned, the company also introduced Epoch Builder Edition. It's a platform designed to help developers, researchers and enterprises build, fine-tune and deploy AI models for Indian use cases. The platform includes new multilingual large language models with 7 billion and 70 billion parameters that support more than 10 Indian languages, including Hindi, Kannada, Tamil, Marathi and Bengali. It also provides GPU infrastructure, training tools, curated datasets, deployment options and safety testing features. Sarvam said the platform is aimed at banks, government departments, startups and research institutions that want to build AI applications using their own data. The company also announced partnerships with three IITs and two state governments for pilot projects in education and public service delivery. Epoch Builder Edition will reportedly enter private preview in August 2026, with a wider rollout planned in the fourth quarter of the year. Also read: Samsung Galaxy S26 FE may bring 45W fast charging, Exynos 2500 chip: Here is what latest leaks reveal Devendra Chaplot joins Sarvam Lastly, Sarvam said that the AI researcher Devendra Singh Chaplot has joined the company as a part-time advisor. Chaplot most recently worked at Elon Musk's xAI and has also been part of the founding teams at Mistral AI and Mira Murati's Thinking Machines Lab. Addressing the audience, Chaplot said India needs AI models that can understand the country's languages and cultural diversity. He added that while AI models may become outdated over time, the ability to build and train them remains valuable. Based in Palo Alto, Chaplot will work with Sarvam's new foundational AI research lab in San Francisco.