State-of-the-art open source models. Covered by BAA.
Get HIPAA-compliant inference for the latest frontier open-source models - at a fraction of the cost of OpenAI/Anthropic etc.
- Cheaper than closed-source on Azure/Bedrock
- ~10xCheaper than closed-source on Azure/Bedrock
- Open source models available now
- 12+Open source models available now
- Unified API - OpenRouter-style routing
- 1Unified API - OpenRouter-style routing
- For enterprise-grade HIPAA-compliance
- BAAFor enterprise-grade HIPAA-compliance
Model catalog
The latest open source models - not on Azure or Bedrock
One API, every model. Swap models with a single parameter, the same way you would with OpenRouter. Run the latest open source models behind HIPAA-compliant infrastructure.
DeepSeek-V4.1-Flash
$0.30/M Input • $1.20/M Output • 1048576 Context
View details →GLM-5.3
$1.4/M Input • $4.4/M Output • 1048576 Context
View details →GLM-5.3-Flash
$0.15/M Input • $0.50/M Output • 1048576 Context
View details →Kimi K3
$3/M Input • $15/M Output • 1048576 Context
View details →
MiniMax M3
$0.3/M Input • $1.2/M Output • 512000 Context
View details →DeepSeek-V4-Pro
$1.32/M Input • $3.96/M Output • 1048576 Context
View details →Kimi K2.7 Code
$0.95/M Input • $4/M Output • 262144 Context
View details →
MiniMax M2.7
$0.3/M Input • $1.2/M Output • 196608 Context
View details →DeepSeek-V4-Flash-Vision-Exp
$0.14/M Input • $0.28/M Output • 1048576 Context
View details →Qwen3.8-Max
$2/M Input • $6/M Output • 262144 Context
View details →Qwen3.8-27B
$0.80/M Input • $4/M Output • 262144 Context
View details →Qwen3.8-Flash-Next
$0.15/M Input • $0.47/M Output • 262144 Context
View details →
Cost comparison
Closed-source on Azure & Bedrock costs up to ~10x more
Open source models are served on optimized inference infrastructure, and the savings are passed through to you.
Closed-source on Azure OpenAI / Bedrock
Indicative cost baseline
- Ecosystem built to push expensive, proprietary models
- Limited open-source model catalog (no access to GLM-5.3, DeepSeek V4, etc.)
- Rigid vendor lock-in to their specific cloud infrastructure
Open source models, unified API
Pass-through optimized inference pricing
- Frontier-level intelligence at a fraction of proprietary model costs
- Unlock access to the world's best open-source models
- Flexible routing commitment spent dynamically across any model or provider
How it works
From waitlist to production in three steps
A single integration gets you from prototype to HIPAA-compliant production inference - with the model flexibility healthcare teams need.
One unified API
OpenRouter-style routing across the full open source catalog. Change models with a single parameter - no re-architecting when something better ships.
HIPAA-compliant by default
AES-256 encryption in transit and at rest. Isolated inference, access controls, and audit logging. Enterprise-grade AI governance.
Deploy & scale
Fast deployment across your organisation with the OpenAI SDK. Scale effortlessly with the one API for any model. Keep up to date with the latest models - no vendor lock-in.
Python
import osfrom openai import OpenAIclient = OpenAI( api_key=os.environ.get("OPENMEDROUTER_API_KEY"), base_url="https://api.openmedrouter.com/inference/v1")stream = client.chat.completions.create( model="glm-5.3", messages=[{"role": "user", "content": "Write a discharge summary for this patient"}], stream=True)for chunk in stream: if chunk.choices[0].delta.content: print(chunk.choices[0].delta.content, end="")Compliance & trust
Built for healthcare
Enterprise HIPAA Compliance
Sign a comprehensive BAA with confidence. Protect patient data (PHI) via zero-retention isolated inference, end-to-end AES-256 encryption (in transit and at rest), and strict least-privilege access controls.
Continuous Audit & AI Governance
Maintain full visibility with real-time access logs and cryptographic inference audit trails. Configure custom retention windows to cleanly support your institutional security reviews and compliance postures.
FAQ
HIPAA, BAAs, and open source models - answered
OpenAI offers a Business Associate Agreement (BAA) for its API under Enterprise terms, but the latest open source models (GLM-5.3 and GLM-5.3-Flash, DeepSeek-V4-Pro and V4-Flash, Kimi K3, Qwen3.8 27B/Flash-Next/Max, and others) are not available through OpenAI at all. OpenMed Router brings those same open source models under a HIPAA-compliant deployment with a signed BAA.
Yes - AWS Bedrock can be configured as HIPAA-eligible and offers a BAA. However, Bedrock supports only a limited subset of models (largely Anthropic, Meta, and Amazon), and closed-source model pricing is often around 10x higher than running equivalent open source models. OpenMed Router gives you a wider catalog of the latest open source models at a fraction of the cost.
Yes. The open source licenses for these models permit commercial use. HIPAA compliance depends on how the model is hosted and whether PHI is protected with a signed BAA, encryption in transit and at rest, and access controls - all of which OpenMed Router provides.
OpenRouter (the popular inference aggregator) is not positioned as a HIPAA-compliant provider and does not provide a BAA for open source model workloads. Here at OpenMed Router, we follow the same unified-API model as OpenRouter, but add HIPAA-compliant infrastructure and an AI Governance layer for covered entities and their business associates.
Closed-source frontier models carry large per-token margins. Open source models can be served efficiently on optimized inference infrastructure, and those savings are passed through. You also avoid cloud egress lock-in and minimum commitments. Exact savings depend on the model and volume you select.
We are currently onboarding design partners, with general availability is coming soon! Join the waitlist to be prioritized - we'll reach out with a qualification call and early access.
Join the waitlist
Be first in line for HIPAA-compliant open source inference
We are currently onboarding design partners, with general availability is coming soon! Join the waitlist to be prioritized - we'll reach out with a qualification call and early access.
By joining the waitlist you agree to be contacted by OpenMed Router about the design partner program.