EpistemeAI/OpenMedResearch-Gemma-4E4N

VISIONConcurrency Cost:1Model Size:7.9BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:Jun 15, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

EpistemeAI/OpenMedResearch-Gemma-4E4N is a 7.9 billion parameter multimodal Transformer model, fine-tuned from google/gemma-4-E4B, designed for open biomedical research. It specializes in biomedical question answering, scientific literature reasoning, and PubMed-style paper search, leveraging the jmhb/PaperSearchQA dataset. This model excels at evidence-grounded research assistant workflows and retrieval-augmented medical research experiments, supporting text, image, and audio inputs with a 128K token context length.

Loading preview...

OpenMedResearch-Gemma-4E4N: Biomedical Research Assistant

EpistemeAI/OpenMedResearch-Gemma-4E4N is a 7.9 billion parameter multimodal model, fine-tuned from google/gemma-4-E4B using the jmhb/PaperSearchQA dataset. This model is specifically engineered for biomedical research and development use only, focusing on reasoning over scientific literature rather than providing direct clinical advice.

Key Capabilities

  • Biomedical Question Answering: Designed to answer factual biomedical questions by reasoning over scientific literature.
  • Scientific Literature Exploration: Facilitates PubMed-style paper search, research summarization, and hypothesis exploration.
  • Retrieval-Augmented QA: Optimized for workflows where retrieved PubMed abstracts or scientific passages provide context for answer generation.
  • Multimodal Support: Based on Gemma 4 E4B, it supports text, image, and audio inputs, generating text outputs, with a context length of up to 128K tokens.

Intended Use Cases

  • Biomedical research question answering
  • PubMed-style scientific paper search
  • Evidence-grounded research assistant workflows
  • Medical and biological factoid QA
  • Scientific search-agent experimentation

Important Limitations and Safety

This model is not for direct clinical diagnosis, treatment planning, or emergency medical decisions. Outputs are preliminary research assistance and require independent verification by qualified professionals. It is aligned to distinguish research information from clinical advice, cite evidence, and express uncertainty, refusing unsafe medical or biological instructions.

Finetunes
2 models