SuperAPI AI Directory

WiseDiag - Hangzhou Zhizhen Technology Launches the World's Leading Medical Large Model

Author: AI小集 · Original source: AI工具集

What is WiseDiag

WiseDiag is the world's leading medical large model launched by Hangzhou Zhizhen Technology, focusing on improving the efficiency and quality of medical services based on artificial intelligence technology. The model has 73 billion parameters and a 32k context length, and is trained on over 3 trillion tokens of professional medical data, covering 10,000 medical professional books, 50,000 clinical diagnosis and treatment guidelines, and 500,000 global medical treatises. Its core functions include deep reasoning capabilities, able to simulate the clinical thinking patterns of expert doctors and conduct health consultations for multi-step complex scenarios. WiseDiag has a long-term memory system that can store users' health data and past medical history, providing personalized suggestions in future consultations. Through the "medical expert avatar" technology, WiseDiag can digitize experts' knowledge and experience, achieving 7×24-hour professional health consultation. WiseDiag has launched multiple versions, including the standard version, enhanced version, and quantized version, respectively suitable for health consultation scenarios of different complexity.

WiseDiag

WiseDiag's Model Versions

WiseDiag-Z1

  • It is the standard version of WiseDiag, with 73 billion parameters and a 32k context length, and a knowledge cutoff date of June 2024. It is trained on massive medical data, covering 10,000 medical professional books, 50,000 clinical diagnosis and treatment guidelines, and 500,000 global medical treatises. It is suitable for most health consultation scenarios and can provide accurate interpretation of medical reports and personalized health suggestions.

WiseDiag-Z1 Thinking

  • It is the deep thinking version, with deep logical reasoning capabilities and a knowledge cutoff date of the end of 2024. It can simulate the clinical thinking patterns of expert doctors and conduct in-depth analysis and repeated weighing in multi-step complex scenarios. It is suitable for complex health consultations and difficult case analysis, and performs outstandingly in the diagnosis of rare diseases and complex conditions.

WiseDiag-Z1 Lite

  • It is the quantized version, with faster response speed and lower cost. While maintaining core functions, it optimizes performance and is suitable for simple, inclusive health consultation scenarios. It is applicable to basic health consultation and rapid test report interpretation.

WiseDiag-Genetics

  • It is a model focused on the field of genetics, able to assist doctors in genetic etiology analysis, recommend testing plans, formulate treatment plans, and provide reproductive guidance. It is mainly used in genetic disease consultation and assisted reproduction, and can also provide learning assistance for genetic counselors.

WiseDiag's Main Functions

  • Deep Reasoning and Diagnosis: WiseDiag has powerful deep reasoning capabilities, able to simulate the clinical thinking patterns of expert doctors and conduct health consultations for multi-step complex scenarios. Through a multi-level reasoning mechanism, it deeply analyzes causes, performs differential diagnosis, and formulates optimal treatment plans.
  • Medical Report Interpretation: WiseDiag can quickly and accurately interpret physical examination reports and test reports, and provide professional advice. It can point out abnormal indicators and, combined with the user's historical health data, provide personalized health consultation reports.
  • Medical Expert Avatar System: Through the "medical expert avatar" technology, WiseDiag can digitize the experience of top experts and achieve 7×24-hour online health consultation. Users can obtain professional doctor-level advice at any time.
  • Long-Term Memory System: WiseDiag has a three-layer memory architecture (real-time memory, short-term memory, and long-term memory), can store users' health data and past medical history, and provide personalized suggestions in future consultations.
  • Open API Platform: WiseDiag provides an API platform for developers and medical institutions to integrate, jointly promoting the development of medical AI.

WiseDiag's Technical Principles

  • Powerful Medical Knowledge Base: WiseDiag is built on 73 billion parameters and a 32k context length. Its training data includes 3 trillion tokens of professional medical data, covering authoritative medical textbooks, clinical guidelines, and high-level medical research papers. The data underwent supervised fine-tuning and direct preference optimization, ensuring the model's professionalism and reliability.
  • Accurate Interpretation of Medical Reports: WiseDiag can quickly and accurately interpret complex medical reports, providing analysis of key indicators and personalized health recommendations. Through medical coding models and multidimensional data analysis, it reproduces doctors' diagnostic thinking and helps patients better understand the content of reports.

WiseDiag's official website address

WiseDiag's model pricing

  • WiseDiag-Z1: Input: ¥0.0125/1k tokens; Output: ¥0.0500/1k tokens.
  • WiseDiag-Z1 Thinking: Input: ¥0.0250/1k tokens; Output: ¥0.1000/1k tokens.
  • WiseDiag-Z1 Lite: Input: ¥0.0100/1k tokens; Output: ¥0.0400/1k tokens.
  • WiseDiag-Genetics: Input: ¥0.0500/1k tokens; Output: ¥0.2000/1k tokens.

WiseDiag's application scenarios

  • Health Management: Provide users with personalized health management plans, supporting chronic disease management and potential risk prediction.
  • Medical Assistance: Help doctors analyze difficult cases, optimize examination plans, and reduce unnecessary medical expenses.
  • Telemedicine: Through AI technology, enable high-quality medical resources to reach remote areas.