Sarvam AI, a Bengaluru-based startup, announced its biggest expansion at its Epoch 2026 event, stating it is building a one-trillion-parameter foundation model from scratch in India. Co-founder Pratyush Kumar said the model will target coding, cybersecurity, scientific research and simulation workloads and will compete with leading global AI systems. The company also announced pricing for its Sarvam 105B model at $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, claiming its offering is about 5.5 times cheaper than GPT-5.4 Mini and more than 11 times cheaper than Gemini 3.5 Flash. Sarvam launched Sarvam Vision Edge, a document intelligence platform, and Sarvam Vision 2.0, an improved optical character recognition; Kumar noted the Odisha government is already using the technology to digitise land records. The company introduced Epoch Builder Edition, a platform for building, fine-tuning and deploying AI models for Indian use cases, including new multilingual large language models with 7 billion and 70 billion parameters supporting more than 10 Indian languages. Sarvam also hired AI researcher Devendra Singh Chaplot, who most recently worked at Elon Musk’s xAI.