Rumiii/LlamaTron-RS1-Nemesis-1B

TEXT GENERATIONPricing:Input $0.108 / Output $0.804Concurrent Unit Cost:1Model Size:1BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Feb 19, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Rumiii/LlamaTron-RS1-Nemesis-1B is a 1 billion parameter medical reasoning model fine-tuned from Meta's Llama-3.2-1B-Instruct. It specializes in complex clinical questions, providing structured and coherent reasoning across differential diagnosis, treatment planning, pharmacology, and clinical case analysis. Trained on 204,773 clinical reasoning conversations, this model is optimized for medical SFT tasks.

Loading preview...

Model Overview

LlamaTron-RS1-Nemesis-1B is a 1 billion parameter language model developed by Rumiii, specifically fine-tuned for medical reasoning tasks. It is based on Meta's Llama-3.2-1B-Instruct architecture and was trained using QLoRA on the extensive Medical-Reasoning-SFT-MiniMax-M2.1 dataset.

Key Capabilities

  • Specialized Medical Reasoning: Excels in handling complex clinical questions, providing structured and coherent reasoning.
  • Comprehensive Clinical Coverage: Trained on a dataset encompassing differential diagnosis, treatment planning, pharmacology, and clinical case analysis.
  • Efficient Performance: Despite its compact 1 billion parameter size, it demonstrates strong capabilities in medical SFT (Supervised Fine-Tuning) scenarios.

Training Details

The model was fine-tuned on the Medical-Reasoning-SFT-MiniMax-M2.1 dataset, which includes 204,773 clinical reasoning conversations with full chain-of-thought traces. Training loss consistently decreased, indicating no overfitting. The model was trained with a maximum sequence length of 512 tokens.

Intended Use and Limitations

This model is designed for research and educational purposes only in the medical domain. It is not intended to be a substitute for professional medical advice, diagnosis, or treatment. Users should be aware that its performance on longer clinical texts may be limited due to the 512-token training sequence length.