MedPsy-4B is a state-of-the-art, text-only medical and healthcare language model purpose-built for edge deployment. Built on top of Qwen3-4B-Thinking-2507 and post-trained with a multi-stage pipeline (supervised fine-tuning + reinforcement learning) on curated medical data, it surpasses models nearly 7x its size on medical benchmarks.
Surpasses 27B models at 4B scale: Scores 70.54 on closed-ended medical benchmarks, outperforming MedGemma-27B-text-it (69.95) despite being nearly 7x smaller
Real-world clinical strength: Achieves 74.00 on HealthBench and 58.00 on HealthBench Hard, beating MedGemma-27B (65.00 / 42.00) by +9.00 and +16.00 points
3.2x token efficiency: Produces accurate medical answers in ~909 tokens vs ~2,953 for Qwen3-4B Thinking, reducing latency and compute cost
Privacy-first: Enables fully on-device inference via the QVAC SDK and QVAC
Fabric, patient data never leaves the device.
MedPsy 4B: Benchmarks
## Benchmark Results
MedPsy-4B
MedGemma-27B-text-it
Qwen3-4B-Thinking-2507
MedGemma-1.5-4B-it
Closed-Ended Medical Benchmarks
Average
70.54
69.95
63.10
51.20
MMLU (Health)
89.70
90.48
85.92
67.69
AfriMedQA
71.50
73.07
64.12
54.38
MMLU-Pro Health
70.45
72.94
67.73
47.31
MedMCQA
72.15
72.77
61.78
50.08
MedQA (USMLE)
84.39
83.29
70.91
64.39
MedXpertQA
30.61
25.18
16.69
15.80
PubMedQA
75.00
71.93
74.53
58.73
HealthBench
Overall
74.00
65.00
63.00
54.00
Expertise-Tailored Communication
79.33
73.00
71.00
62.67
Response Depth
63.67
61.33
58.00
48.67
Context Seeking
71.67
58.67
57.67
46.00
Emergency Referrals
81.67
73.00
74.00
64.00
Global Health
73.67
61.00
59.00
47.67
Health Data Tasks
60.67
56.67
54.67
44.67
Responding Under Uncertainty
76.33
66.33
64.33
58.33
HealthBench Hard
Overall
58.00
42.00
42.67
29.67
Expertise-Tailored Communication
55.33
44.67
45.00
31.67
Response Depth
47.67
38.67
38.67
29.00
Context Seeking
63.33
42.00
43.00
28.00
Emergency Referrals
62.33
39.67
47.33
29.00
Global Health
60.00
42.67
43.33
29.00
Health Data Tasks
46.67
39.33
39.67
23.67
Responding Under Uncertainty
61.00
42.67
42.00
35.00
* MMLU (Health): averaged accuracy across 6 sub-domains: anatomy, clinical_knowledge, college_biology, college_medicine, medical_genetics, professional_medicine.
* HealthBench evaluated using CompassJudger-2-32B-Instruct as judge.
* All results are averaged over 3 runs with generation parameters: temperature=0.6, top_k=20, top_p=0.95, max_output_tokens=16384.
Token Efficiency
Beyond raw accuracy, MedPsy-4B achieves a 3.2x reduction in average response length compared to its backbone model (Qwen3-4B-Thinking-2507). This means faster inference, lower compute costs, and reduced latency - critical for real-time clinical decision support on edge devices.
Qwen3-4B-Thinking-2507
MedPsy-4B
Avg. Response Length (Tokens)
2,953
909
Δ Reduction
3.2x fewer tokens
The chart below shows per-benchmark response lengths. The largest reductions appear on reasoning-intensive tasks (MedXpertQA, MedQA-USMLE, MMLU-Pro Health), where the base model's extended thinking produces substantially longer outputs without accuracy gains over our post-trained model.
Average Response Length (Tokens) - 4B model class
Average response length (tokens) per benchmark. Lower is better. MedPsy-4B consistently produces shorter responses than Qwen3-4B-Thinking-2507 while achieving higher overall accuracy.
Model Details
Parameter
Value
Architecture
Qwen3ForCausalLM
Parameters
4B
Hidden size
2,560
FFN hidden size
9,728
Layers
36
Attention heads
32
KV groups (GQA)
8
Vocab size
151,936
Max position embeddings
262,144
Precision
bfloat16
Position embedding
RoPE
Normalization
RMSNorm
Activation
SwiGLU
Usage
Transformers
python
1from transformers import AutoModelForCausalLM, AutoTokenizer
23model_name ="qvac/MedPsy-4B"45tokenizer = AutoTokenizer.from_pretrained(model_name)6model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype="auto", device_map="auto")78messages =[9{"role":"user","content":"What are the common symptoms and first-line treatments for community-acquired pneumonia?"}10]1112text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)13inputs = tokenizer([text], return_tensors="pt").to(model.device)1415outputs = model.generate(**inputs, max_new_tokens=1024)16response = tokenizer.decode(outputs[0][inputs.input_ids.shape[-1]:], skip_special_tokens=True)17print(response)
Training
The model was post-trained through a multi-stage pipeline on the Qwen3-4B-Thinking-2507 backbone:
SFT Stage 1 (Corpus 1): Broad medical adaptation on a large-scale synthetic corpus spanning biology, medicine, and health (including a new health domain not yet publicly released), built from Genesis II–style medical seeds and open-source medical QA prompts used purely as questions, with all reasoning targets freshly generated by Baichuan-M3-235B.
SFT Stage 2 (Corpus 2): Reasoning specialization on a smaller, higher-value clinical QA corpus with teacher-generated chain-of-thought reasoning from Baichuan-M3-235B.
RL Stage 1: Reinforcement learning (DAPO) on the easy/moderate subset of AlphaMedQA (Liu et al., 2025), annotated with the SFT checkpoint.
RL Stage 2: Focused RL on a hard-enriched AlphaMedQA subset re-annotated with the best Stage 1 checkpoint, targeting persistent failure modes.
MedPsy-4B is an open language model intended as a starting point for developers and researchers building downstream healthcare applications involving medical text. Developers are expected to validate, adapt, and make meaningful modifications to the model for their specific use cases.
Appropriate use cases include:
Research on medical language understanding and reasoning
Building developer tools and prototypes for health-related applications
On-device medical information retrieval for privacy-sensitive environments
Always with appropriate disclaimers.
Limitations
[!WARNING]
This model is NOT a substitute for professional medical judgment and the model outputs are NOT a substitute for proper clinical diagnosis. Always consult with a certified physician. Despite strong benchmark performance, MedPsy-4B is a compact 4B-parameter language model that will make errors. Medical AI systems can produce outputs that appear confident and authoritative while being factually incorrect, incomplete, or clinically inappropriate.
Known limitations include:
Hallucinations: The model may generate plausible-sounding but incorrect medical information.
English only: The model was trained and evaluated primarily in English. Performance in other languages is not validated.
Text only: This model processes text inputs only. It cannot interpret medical images, lab results in non-text formats, or other modalities.
No real-time knowledge: The model's knowledge has a training data cutoff and does not reflect the latest medical guidelines, drug approvals, or clinical evidence.
Bias in training data: As with any model trained on synthetic and public medical data, biases in the source material may propagate to model outputs. Developers should validate performance across diverse patient populations, demographics, and clinical contexts.
Not designed for emergencies: This model should never be used as the sole decision-making tool in emergency or life-threatening situations.
Safety Recommendations
When integrating this model into any application:
Always include visible disclaimers informing users that outputs are AI-generated and not a substitute for professional medical advice
Do not use for direct clinical diagnosis or treatment without oversight by qualified healthcare professionals
Monitor for harmful outputs and implement appropriate safety filters in production systems
Ethics and Safety
The model was evaluated on medical safety dimensions through the HealthBench evaluation framework, which assesses Emergency Referrals, Responding Under Uncertainty, and Context Seeking, all critical safety dimensions for medical AI. However, no dedicated red-teaming or adversarial safety testing has been conducted on this model to date. Developers deploying this model in production should conduct their own safety evaluations appropriate to their use case.
Related Resources
MedPsy Collection: All MedPsy models, datasets, and resources in one place
1@article{medpsy2026,
2 title={MedPsy: State-of-the-Art Medical and Healthcare Language Models for Edge Devices},
3 author={Vitabile, Davide and Buffa, Alexandro and Nambiar, Akshay and Nazir, Amril},
4 year={2026},
5 url={https://huggingface.co/blog/qvac/medpsy}
6 institution={Tether AI Research}
7}
Copyright
We will take appropriate actions in response to notices of copyright infringement. If you believe your work has been used or copied in a manner that infringes upon your intellectual property rights, please email data-apps@tether.io identifying and describing both the copyrighted work and alleged infringing content.
As described above, a subset of the Genesis I and Genesis II datasets was used by the Baichuan-M3-235B model—which itself is also available under the Apache 2.0 license to generate synthetic data for training this model. The Genesis I dataset is made available under the CC-BY-NC 4.0 (Creative Commons Attribution-NonCommercial 4.0) license. The Genesis II dataset is also made available under the CC-BY-NC 4.0 license.
