9 Downloads Updated 1 month ago
ollama run ktiyab/coheara-medgemma-4b-f16
Full-precision variant of MedGemma 1.5 4B IT for local medical document understanding. Built from safetensors for supply chain integrity — not from community GGUFs.
| Base model | google/medgemma-1.5-4b-it |
| Quantization | F16 (no quantization — full precision) |
| Size | 8.6 GB |
| Context | 8192 tokens |
| Modalities | Text + Vision (multimodal) |
| Built by | Coheara from original safetensors |
MedGemma is a collection of Gemma 3 variants fine-tuned by Google for performance on medical text and image comprehension. MedGemma 1.5 4B is an instruction-tuned, decoder-only multimodal Transformer designed to accelerate building healthcare-based AI applications.
| Specification | Details |
|---|---|
| Model type | Decoder-only Transformer (Gemma 3) |
| Parameters | 4B (multimodal) |
| Attention | Grouped-query attention (GQA) |
| Image encoder | SigLIP (pre-trained on de-identified medical data) |
| Image resolution | 896 × 896 (normalized), 256 tokens per image |
| Max output | 8192 tokens |
| Benchmark | Score |
|---|---|
| MedQA (4-op) | 69.1% |
| MedMCQA | 59.8% |
| MMLU Med | 69.6% |
| EHRQA | 89.6% |
| MIMIC CXR (Macro F1 top 5) | 89.5% |
Source: MedGemma Technical Report (Sellergren et al., 2025)
# Text query
ollama run coheara-medgemma-4b-f16 "What are the common side effects of metformin?"
# Vision — analyze a medical document
ollama run coheara-medgemma-4b-f16 "What medications are prescribed in this document?" --images ./prescription.jpg
import ollama
# Text query
response = ollama.chat(
model="coheara-medgemma-4b-f16",
messages=[{
"role": "user",
"content": "What are the common side effects of metformin?"
}]
)
print(response["message"]["content"])
# Vision — medical document extraction
with open("prescription.jpg", "rb") as f:
image_data = f.read()
response = ollama.chat(
model="coheara-medgemma-4b-f16",
messages=[{
"role": "user",
"content": "What medications are prescribed? For each, state the name, dose, and instructions.",
"images": [image_data]
}]
)
print(response["message"]["content"])
curl http://localhost:11434/api/chat -d '{
"model": "coheara-medgemma-4b-f16",
"messages": [{
"role": "user",
"content": "What are the common side effects of metformin?"
}]
}'
import ollama
stream = ollama.chat(
model="coheara-medgemma-4b-f16",
messages=[{
"role": "user",
"content": "Explain the mechanism of action of ibuprofen."
}],
stream=True
)
for chunk in stream:
print(chunk["message"]["content"], end="", flush=True)
Coheara is a local, offline, encrypted desktop application that serves as a patient’s personal medical AI companion. It runs entirely on the user’s machine — no cloud, no data leaving the device.
MedGemma 1.5 4B is Coheara’s core extraction engine. It reads medical documents (prescriptions, lab reports, clinical letters) through vision and extracts structured information that the application stores in an encrypted local database.
We produce multiple quantized variants from Google’s original safetensors to match different hardware capabilities:
| Variant | Quantization | Size | Target Hardware |
|---|---|---|---|
| q4s | Q4_K_S | 3.2 GB | Smallest supported. Research only. |
| q4 | Q4_K_M | 3.3 GB | Low RAM/VRAM (≥4 GB) |
| q8 | Q8_0 | 5.0 GB | Balanced (≥8 GB) — recommended |
| f16 | F16 | 8.6 GB | ← This model. Full precision (≥12 GB) |
All variants are built directly from google/medgemma-1.5-4b-it safetensors using ollama create. The F16 variant preserves the original weight precision — no quantization is applied. This matters because community-converted GGUFs for Gemma 3 can break the vision tower (ollama/ollama#9967).
Important: F16 has not been directly benchmarked in Coheara’s extraction pipeline. The benchmark data below is inferred from the quantized variants and Google’s published results. F16 preserves the original model weights at full precision — it is expected to perform equal to or better than all quantized variants on every metric.
These benchmarks reflect Coheara’s specific extraction workload, not general medical AI performance. They are not comparable to Google’s published MedQA/MMLU scores.
The quantization ladder establishes a clear pattern: lower precision increases degeneration risk under complex prompts. F16 sits at the top — maximum precision, minimum degeneration risk.
| Strategy | Q4_K_S CPU | Q4_K_M CPU | Q8_0 CPU | F16 CPU (expected) |
|---|---|---|---|---|
| All-at-once JSON (BM-04) | Not tested | 36% degen | 0% degen | 0% degen |
| Markdown list (BM-05) | 0% degen | 0% degen | Expected 0% | 0% degen |
| Iterative drill (BM-06) | 25% degen | 0% degen | Expected 0% | 0% degen |
For most users, coheara-medgemma-4b-q8 (5.0 GB) provides equivalent reliability at nearly half the size. Q8 achieves 0% degeneration on CPU for all strategies, and the 3.6 GB size difference is significant for local deployment. F16 is primarily for users with hardware headroom who want to eliminate any quantization-related uncertainty.
This model is a Model Derivative of MedGemma 1.5 4B IT, which is released under Google’s Health AI Developer Foundations Terms of Use. By downloading or using this model, you agree to be bound by these terms.
This model is not approved for direct clinical use. It is a developer tool — a starting point for building healthcare applications that must undergo their own validation, regulatory review, and clinical evaluation before deployment.
No direct clinical use without validation. You must not use this model to directly inform clinical diagnosis, patient management decisions, or treatment recommendations without appropriate validation, adaptation, and regulatory authorization for your specific use case.
Regulatory authorization required. If your application falls within the scope of health regulatory oversight (e.g., as a medical device), you must seek Health Regulatory Authorization before deployment.
Pass-through restrictions. If you distribute this model or any derivative of it, you must include the HAI-DEF use restrictions as an enforceable provision in your agreement with recipients, and provide them with the full HAI-DEF Terms of Use.
Modified files notice. Modified files must carry prominent notices stating that you modified them. Although this variant preserves F16 precision, it is repackaged with a Gemma 3 chat template via ollama create and constitutes a Model Derivative.
Prohibited uses. This model must not be used for unlicensed medical practice, generation of misleading health information, or any purpose listed in Google’s Prohibited Use Policy. This includes but is not limited to:
Indemnification. You agree to defend and indemnify Google against all liabilities arising from your use of this model or violation of these terms.
You can use this model to build and evaluate healthcare AI applications. You cannot use it as-is to make clinical decisions about real patients. If you share it, you must pass along these same restrictions. If you deploy it in a regulated context, you must get regulatory approval first.
Full terms: developers.google.com/health-ai-developer-foundations/terms
| Source | google/medgemma-1.5-4b-it (safetensors, gated access via HuggingFace) |
| Build method | ollama create from original weights (F16, no quantization applied) |
| Chat template | Gemma 3 (<start_of_turn>user/model, <end_of_turn>) |
| No community GGUFs | Built from safetensors to preserve vision tower integrity |
If you use this model in research, please cite both the original MedGemma paper and this variant:
@article{sellergren2025medgemma,
title={MedGemma Technical Report},
author={Sellergren, Andrew and Kazemzadeh, Sahar and others},
journal={arXiv preprint arXiv:2507.05201},
year={2025}
}
This model is subject to the Health AI Developer Foundations Terms of Use.