OpenMed

OpenMed is an open-source clinical AI library that reads medical text and removes protected health information on local hardware.

OpenMed is a local-first clinical AI library that reads clinical text and removes over 55 types of protected health information on hardware you control. It features more than 2,000 open-source models, language-aware support across 35 PII languages, and reproducible biomedical named entity recognition benchmarks. The software runs across multiple runtimes including Apple MLX, PyTorch, ONNX Runtime Mobile, and WebGPU, operating entirely offline without cloud APIs or telemetry. It supports structured outputs, REST and gRPC services, context-aware PII detection, format validators for national identifiers, and five redaction methods. Designed for clinical teams, researchers, and mobile developers, OpenMed provides public weights, training recipes, and evaluation scripts for full auditing and fine-tuning.

OpenMed at a glance

Best for: Researcher, Software engineer
Runs on: Web, macOS, Windows, Linux, iOS, Android, API, self-hosted, open source
Pricing: Free
Released: October 2025
Category: AI research tool
Website: openmed.life

Key features

Removes 55+ PHI types on local hardware
Supports 35 model-backed PII languages
Offers REST and gRPC endpoints with built-in auth
Performs context-aware PHI detection and validation
Provides five distinct redaction methods
Runs on CPU, Apple Silicon, NVIDIA GPU, mobile, and browser
Includes specialized models across 13 biomedical domains

Use it for

De-identifying clinical notes locally before submission
Extracting biomedical entities for research and discovery
Running offline clinical AI on mobile devices
Fine-tuning biomedical models on custom vocabularies

Alternatives to OpenMed

OpenResearch — Run parallel research workflows locally or on remote compute.
Open Science — Local-first design with persistent Python and R kernels and traceable artifacts.
OpenClaw Medical Skills — Adds 869 clinical, genomic, and drug discovery skills to OpenClaw and NanoClaw agents.
Open Health Sync — It normalizes thirteen health data sources into one clean model with a single set of units.
Biomni — Automates complex wet-lab experimental design with specialized tools and databases.
OpenRadView — Explores local medical scans without accounts, telemetry, or cloud uploads.
OpenCompass — It offers an objective, comprehensive, and credible open-source AI evaluation system.
Deepopen — Provides millisecond-level inference speeds and completely eliminates text generation hallucinations.
Morphik — Provides native Model Context Protocol support and granular multi-user access.
Scite — It uses direct publisher agreements to verify scientific claims with exact source links.
OpenOSINT — Provides 20 automated intelligence tools with natural-language AI agent interfaces.
ASReview — It reduces the manual workload of large-scale text screening by up to 95%.

All OpenMed alternatives