- On the Open Medical LLM Leaderboard Jivi MedX First place worldwide, GPT-4 and Med-PaLM2 also held back
- Jivi aims to launch its AI healthcare product by the end of this year
India A purpose-built medical large language model (LLM) developed by Jivi, an Indian healthcare AI startup co-founded by Ankur Jain, former chief product officer of BharatPay, and GV Sanjay Reddy, chairman of Reddy Ventures, has topped the Open Medical LLM leaderboard. Jivi's LLM Jivi MedX outperformed established LLMs including OpenAI's GPT-4 and Google's Med-PaLM 2 with an average score of 91.65 across the leaderboard's nine benchmark categories. Indian AI Startup Jivi Emerges as World's Number 1, Beating OpenAI and Google
A leaderboard hosted by leading AI platform Hugging Face, University of Edinburgh and Open Life Science AI ranks medical-related LLMs based on their performance in answering medical questions from exams and research. The assessment covers medical exams like India's Medical Entrance Exam (AIIMS and NEET), US Medical Licensing Exam (USMLE) and comprehensive assessments in clinical knowledge, medical genetics and professional medicine.
“JIVI is revolutionizing primary healthcare through generative AI, making high-quality care accessible 24/7 at a fraction of the cost. Our mission at JIVI is to use artificial intelligence to enhance patient care. Our platform accelerates diagnosis and ensures high precision enabling timely and accurate treatment for all” Said Ankur Jain, Co-Founder and CEO of JIVI.
JIVI currently works with a 20-member team of physicians, surgeons, AI engineers and data scientists. It is developing technologies to transform the accessibility, affordability and quality of healthcare globally.
“This is another achievement for an Indian company. Our mission at JIVI is to make the best healthcare available to everyone globally. Our LLM being the best worldwide gives us great pride and confidence as we prepare to take JIVI to over a billion people” Said GV Sanjay Reddy, Co-Founder and Chairman of JIVI.
JIVI uses its own vast medical dataset containing millions of medical research papers, journals, clinical notes and other sources to train its JIVI MedX LLM. This dataset is one of the largest datasets in the world. JIVI MedX was trained using an instruction fine-tuning algorithm called Odds Ratio Preference Optimization (ORPO).
