What is the news?
Bengaluru-based artificial intelligence (AI) startup Sarvam AI on Wednesday launched two major language modeling models. Sarvam-30B and Sarvam-105B models have been announced at the India AI Impact Summit. These models are designed for reasoning, multilingual communication, coding, research, and enterprise applications. This is an important step for Indian startups looking to develop domestic AI infrastructure capable of competing with established systems like ChatGPT and other international models.
Sarvam-30B provides better reach
SERVAM-30B is the smaller model of the two, with 30 billion parameters and a context window of 32,000 tokens. It has been trained on 16,000 billion tokens. It is designed for widespread access. In a live demonstration, Sarvam demonstrated a multilingual chatbot called ‘Vikram’, which is capable of carrying on conversations in Hindi, Punjabi, Marathi and other Indian languages, even on feature phones. This shows that AI can reach users beyond smartphones.
This model is useful for industries
The SARVAM-105B model has a context window of 105 billion parameters and 128,000 tokens. It is designed for more complex logic and large-scale analytical tasks. During the summit, the model analyzed a company’s balance sheet in real time and provided accurate answers to detailed financial and contextual questions. Its capabilities make it suitable for enterprise use cases such as analysis, research, coding assistance, and summarization of long documents.
Follow Us On Google News
Follow Us On WhatsApp
Leave a Comment