Hippocratic AI announced the launch of Polaris 5.0, the newest version of its healthcare-focused AI platform designed for voice-based patient interactions. The company said Polaris 5.0 was built using more than 180 million real patient interactions and validated by over 7,500 U.S.-licensed clinicians. Hippocratic AI claims the model outperformed major frontier AI models from OpenAI, Anthropic, and Google on healthcare-specific benchmarks related to clinical accuracy, HIPAA and regulatory compliance, empathy, and conversational consistency. The company described Polaris 5.0 as a 5-trillion-parameter AI constellation powered by a 700-billion-parameter core model, with new capabilities including drug safety support, clinical escalation detection, multilingual conversations, specialty scheduling, and acoustic voice analysis like cough detection.
Hippocratic AI said Polaris 5.0 introduces more than 20 new clinical skills and was specifically designed for safe, real-time healthcare voice interactions. The platform includes features such as multi-document intelligence for insurance and formulary navigation, escalation support across seven body systems, mid-call language switching between English, Spanish, and Mandarin, and improved emotional support performance using the company’s HEART benchmark framework. According to the company, Polaris 5.0 achieved high benchmark scores in areas such as drug safety, HIPAA authentication, CMS guideline adherence, and empathetic patient communication while maintaining low voice-response latency suitable for live patient calls. Hippocratic AI said the release continues its broader mission of developing non-diagnostic, safety-focused generative AI healthcare agents for providers, payers, and pharmaceutical organizations.