Rumiii/Mistral-BioMed-Tool-Caller-7B

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7BQuant:FP8Context Size:4kTool Calling:SupportedPublished:Aug 24, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Rumiii/Mistral-BioMed-Tool-Caller-7B is a 7 billion parameter Mistral-7B-Instruct-v0.3 based model, fine-tuned by Rumiii for biomedical tool-calling. It specializes in selecting and generating properly formatted function-call arguments for biomedical APIs, covering genomics, proteomics, and comparative biology. The model was trained using QLoRA on the BioTool dataset, enabling it to handle 127 tools across NCBI E-utilities, UniProt REST, and Ensembl REST. It maintains general conversational ability while excelling in specialized biomedical function calling.

Loading preview...

Mistral-BioMed-Tool-Caller-7B: Specialized Biomedical Tool-Calling

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.3, developed by Rumiii, specifically designed for biomedical tool-calling. It excels at interpreting biomedical questions and, given a set of tool schemas, accurately selects the appropriate tool and generates correctly formatted function-call arguments in Mistral's native [TOOL_CALLS] format.

Key Capabilities & Features

  • Biomedical Specialization: Trained on 127 tools across three major biomedical API families:
    • NCBI E-utilities: For searching and fetching data from Entrez databases (e.g., PubMed, BLAST).
    • UniProt REST: For protein, proteome, and taxonomy lookups.
    • Ensembl REST: For coordinate mapping, variant effect prediction, and comparative genomics.
  • Tool-Calling Accuracy: Fine-tuned using QLoRA (4-bit) on the gxx27/BioTool dataset, a comprehensive collection of 7,040 human-verified query-to-API-call pairs.
  • Preserved General Ability: A small portion of general-instruction data was included during training to ensure the model retains its base conversational capabilities alongside its new specialization.
  • Efficient Training: Achieved with Unsloth and Hugging Face's TRL library on a single Kaggle Tesla T4, demonstrating efficient resource utilization.

Ideal Use Cases

  • Automated Biomedical Research: Generating API calls for data retrieval and analysis in genomics, proteomics, and comparative biology.
  • Biomedical Question Answering: Translating natural language biomedical queries into executable tool functions.
  • Integration with Biomedical Workflows: Serving as an intelligent agent to interact with NCBI, UniProt, and Ensembl APIs.

Limitations

  • Scope: This model's specialization is limited to genomics, proteomics, and comparative biology tool-use. It does not cover clinical-facing APIs such as ICD-10 lookup, drug databases, or clinical trial registries.