The Engine Showcase

Evaluate our foundation models. Access the registry, run simulated inferences, and review our clinical benchmarks against generic architectures.

Model Registry

ECG / 110M
Arrhythmia Classification

Nano Triage

Ultra-lightweight edge model designed for wearable integration (Apple Watch, Whoop). Capable of realtime classification of 12 common arrhythmias with 99.1% sensitivity.

ID: SFM-ECG-Nano-Triage
Explore
ECG / 1.5B
Comprehensive 12-Lead Analysis

Clinical Compute

Hospital-grade diagnostic engine. Can identify subtle ischemia, structural abnormalities, and predict imminent AFib onset from standard 12-lead clinical ECGs.

ID: SFM-ECG-Clinical-Compute
Explore
Angiogram / 8B
Stenosis Quantification

Vision Assist

Multimodal visual reasoning model for Cath Lab environments. Quantifies plaque burden and stenosis severity in real-time during fluoroscopy.

ID: SFM-Angio-Vision-Assist
Explore

Interactive Playground

Test our inference API directly. The engine returns structured JSON containing diagnostic assessments, confidence intervals, and the critical explainability layer.

ENDPOINT: wss://api.sereneai.in/v1/stream
AUTH:Bearer <ENTERPRISE_KEY>
terminal — bash — 80x24
# SereneAI Clinical API Simulator
~

Performance Benchmarks

Compared against generic multimodal models (GPT-4V, Med-PaLM), our specialized architecture achieves higher clinical accuracy with a fraction of the compute footprint.

SFM-ECG-1.5BF1: 98.2% | 12ms
Generic Multimodal (Cloud)F1: 89.4% | 1450ms
Traditional CNN BaselineF1: 84.1% | 45ms