tokyotech-llm/Medical-Qwen3-Swallow-32B

TEXT GENERATIONPricing:Input $0.408 / Cached $0.0816 / Output $1.972Concurrent Unit Cost:2Model Size:32BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jun 10, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Medical-Qwen3-Swallow-32B is a 32 billion parameter medical-domain language model developed by tokyotech-llm, based on the Qwen3-Swallow architecture. This bilingual Japanese-English model is adapted for medical contexts through continual pre-training and fine-tuning. It is specifically designed to support research and development of safe and trustworthy AI for Japanese clinical settings, demonstrating improved performance on Japanese medical and healthcare-related benchmarks compared to its base model. The model has a context length of 32768 tokens and is intended for research use in medical AI safety and reliability evaluation.

Loading preview...

Medical-Qwen3-Swallow-32B: A Specialized Medical LLM

Medical-Qwen3-Swallow-32B is a 32 billion parameter language model developed by tokyotech-llm, specifically adapted for the medical domain. It builds upon the tokyotech-llm/Qwen3-Swallow-32B-RL-v0.2 base model, inheriting its bilingual Japanese-English capabilities and Qwen3 architecture. This model is designed to advance research and development in safe and reliable AI for Japanese clinical environments.

Key Capabilities and Features

  • Medical Domain Adaptation: Fine-tuned with a mixture of medical-domain text, including biomedical literature, synthetic medical data, and clinical guidelines, while retaining general-domain knowledge.
  • Bilingual Support: Inherits robust Japanese and English language processing from the Qwen3-Swallow family.
  • Enhanced Medical Performance: Demonstrates improved scores across various Japanese medical and healthcare-related benchmarks, such as IgakuQA, JJSIMQA, JMMLU Medical, and MedMCQA_JP, compared to its base model.
  • Research-Oriented: Intended for research and development, particularly for evaluating AI safety and reliability in medical contexts.
  • Standard Compatibility: Compatible with Hugging Face Transformers and vLLM-compatible inference stacks, supporting a context length of up to 32,768 tokens.

Intended Use Cases

  • Medical AI Research: Ideal for researchers exploring the application of large language models in medical and healthcare fields.
  • Safety and Reliability Evaluation: Useful for developing and testing AI systems that require high levels of safety and trustworthiness in clinical settings.
  • Japanese Medical Text Processing: Suited for tasks involving Japanese medical documentation, queries, and information extraction.

Note: This model is for research and development only and has not been validated as a medical device. It should not be used as a substitute for professional medical judgment.