Model Overview

Baichuan-M2-32B is Baichuan AI's second medical-enhanced reasoning model, a 32.8 billion parameter model built on Qwen2.5-32B. It is specifically designed for real-world medical reasoning tasks, incorporating an innovative Large Verifier System and domain-specific fine-tuning. The model aims to provide advanced medical capabilities while maintaining strong general performance.

Key Innovations & Features

Large Verifier System: Combines medical scenario characteristics with a comprehensive medical verification framework, including patient simulators and multi-dimensional verification mechanisms.
Medical Domain Adaptation: Utilizes Mid-Training for lightweight and efficient medical domain adaptation, preserving general capabilities.
Multi-stage Reinforcement Learning: Employs a hierarchical training strategy to progressively enhance medical knowledge, reasoning, and patient interaction.
Leading Medical Performance: Achieves top scores on HealthBench among open-source models, with medical capabilities approaching GPT-5.
Doctor-Thinking Alignment: Trained on real clinical cases and patient simulators to develop clinical diagnostic thinking and robust patient interaction.
Efficient Deployment: Supports 4-bit quantization, enabling deployment on a single RTX4090, and offers 58.5% higher token throughput in MTP version for single-user scenarios.

Performance Highlights

Baichuan-M2-32B demonstrates strong performance across both medical and general benchmarks:

HealthBench: Scores 60.1, outperforming models like gpt-oss-120b and Qwen3-235B-A22B-Thinking-2507.
General Benchmarks: Achieves 83.4 on AIME24, 72.9 on AIME25, and 77.6 on CFBench, showing competitive general reasoning and writing capabilities compared to Qwen3-32B (Thinking).

Usage & Limitations

This model is intended for medical education, health consultation, and clinical decision support. It is crucial to note that it is for research and reference only and cannot replace professional medical diagnosis or treatment. Safe use under the guidance of medical professionals is recommended.