EpistemeAI/Reasoning-Medical0.1-27B

VISIONConcurrency Cost:2Model Size:27BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:Jun 20, 2026License:ccArchitecture:Transformer Cold

EpistemeAI/Reasoning-Medical0.1-27B is a 27 billion parameter causal language model developed by EpistemeAI, fine-tuned for advanced medical reasoning across professional medicine, medical genetics, college biology/medicine, and clinical knowledge. It was trained on 100,000 curated medical reasoning records, incorporating Chain-of-Thought, and supports a native context length of 262,144 tokens, extensible up to 1,010,000 tokens. This model excels in medical question-answering and reasoning tasks, demonstrating high performance on benchmarks like MMLU-Pro Biology (0.93) and MedQA (0.969).

Loading preview...

Model Overview

EpistemeAI/Reasoning-Medical0.1-27B is a 27 billion parameter causal language model developed by EpistemeAI, specifically fine-tuned for advanced medical reasoning. It is designed to excel in professional medicine, medical genetics, college biology/medicine, and clinical knowledge domains.

Key Capabilities & Features

  • Specialized Medical Reasoning: Fine-tuned on 100,000 records from public medical reasoning datasets, incorporating Chain-of-Thought for step-by-step medical question answering.
  • High Performance: Achieves strong benchmark results, including 0.93 on MMLU-Pro Biology, 0.968 on MMLU-ProX Biology, and 0.969 on MedQA, outperforming comparable models.
  • Extended Context Length: Natively supports up to 262,144 tokens, with extensibility up to 1,010,000 tokens using techniques like YaRN for ultra-long text processing.
  • Multimodal Input: Supports text, image, and video inputs, making it versatile for various medical data types.
  • Agentic Usage: Excels in tool-calling capabilities and is compatible with frameworks like Qwen-Agent for building advanced applications.
  • Thinking Mode: Operates in a default 'thinking mode' to generate reasoning content before final responses, with an option to disable for direct answers.
  • Preserved Thinking: Can retain and leverage thinking traces from historical messages, beneficial for agent scenarios to enhance decision consistency and optimize token usage.

Use Cases

  • Medical Education: Assisting with learning and understanding complex medical concepts.
  • Clinical Reasoning Research: Supporting research into diagnostic processes and treatment pathways.
  • Public Health Analysis: Aiding in the analysis of health data and trends.
  • Defensive Biosecurity: Contributing to discussions and research on biosecurity measures.

Important Considerations

  • Safety Notice: Intended for benign medical and scientific reasoning only. Not to be used for harmful activities or direct clinical practice. All outputs require expert review and compliance with safety standards.
  • Performance Optimization: Recommendations for sampling parameters and output length are provided to achieve optimal performance across different tasks.