Healthcare AI Gateway: Top 6 With BAAs Compared (Nov 2026)
Healthcare AI gateways compared: Concentrate covers 7 providers and 70 models under one BAA ($499/month); Aptible covers 3. Plus Bedrock, Azure, Bifrost.
Concentrate.ai
On this page
A healthcare AI gateway is one API that sends your app's requests to LLMs while keeping protected health information (PHI) on routes covered by a Business Associate Agreement (BAA). The question that matters is who signs the BAA for the inference itself. Among the healthcare AI gateways that show up for this search, only two hold BAAs with the model providers and sell you covered inference: Concentrate and Aptible. Concentrate covers 7 providers; Aptible covers 3.
This list was checked against each vendor's own pages in November 2026.
What should a healthcare AI gateway do?
A healthcare AI gateway should let you send PHI to many models under one BAA, and block every route that BAA does not cover. In practice that means five things:
A signed BAA that covers the inference, not just the gateway software.
Routing that only reaches BAA-covered providers, including fallbacks.
Zero data retention at the provider, and no storage of prompts or responses in the gateway's logs.
Several model families (Claude, GPT, Gemini, open models), so you are not stuck with one vendor's catalog.
Controls over who can call which model, with spend limits per key.
Under HHS guidance, a cloud or AI service that handles ePHI for you must sign a BAA. A gateway that only forwards your own provider keys does not change who has to sign.
LLM Gateway
Instant access to every model
Change models, providers, add fallbacks, and more by running all AI usage through one platform.
Why do EHR and agent vendors show up for "healthcare AI gateway"?
Many of the ads on this search are for customer-support agents and EHR integration tools. Those products connect to records systems or answer patients; they do not give your engineers BAA-covered access to Claude, GPT and Gemini through one API. If you are building an AI feature that calls models directly, you need an LLM gateway with a BAA, which is what this list covers.
1. Concentrate: one BAA across 7 providers and 70 models
Concentrate signs one BAA with you and holds BAAs with the providers behind it. Today the BAA-eligible providers are Anthropic, OpenAI, Azure, Google Vertex AI, Fireworks AI, Azure Fireworks and Baseten. That is 70 models and 117 covered routes, including Claude Opus 5.5 and Sonnet 5.5, GPT-6.1 Sol, Gemini 4 Argon, DeepSeek V4.1 Flash, GLM-5.3, Kimi K3 and Qwen3.8 Max. The full catalog is about 215 models from about 30 providers; the covered list is on the HIPAA LLM gateway page.
On a HIPAA organization, zero data retention is locked on, request-content logging is locked off, and uncovered routes are blocked, including fallbacks. If no covered route is left, the request is rejected. The auto router only picks covered routes. Many models have more than one covered route (Claude Sonnet 5.5 runs on Anthropic and Azure), so a provider outage can fail over without leaving the BAA.
Pricing: $499/month with a BAA. Inference is billed at provider rates with no per-request fees and no markup on tokens, so the subscription stays the same however many tokens you send. Without a BAA there is no subscription fee.
Best for: teams that want Claude, GPT, Gemini and open models with PHI under one BAA, with covered failover.
2. Aptible LLM Gateway: one BAA, 3 providers
Aptible also signs a BAA that covers inference. Its own page says it gives access to Anthropic, OpenAI and Amazon Bedrock through one API. Its model list has no Gemini models, and its token prices sit above provider list rates (Claude Opus 5.5 is listed at $4.75 input and $23.77 output per million tokens, against $4 and $20 at provider rates). The gateway is $200/month plus token usage.
Best for: teams already hosting on Aptible that only need Anthropic, OpenAI and Bedrock models. For a side-by-side, see Aptible LLM Gateway vs Concentrate.
3. Amazon Bedrock: AWS models under the AWS BAA
Bedrock is on the AWS HIPAA Eligible Services list, covered by the AWS BAA you accept in AWS Artifact. It covers models served inside AWS only, and you check eligibility model by model. No Gemini or OpenAI frontier models.
Best for: teams fully committed to AWS.
4. Azure AI Foundry: Azure models under Microsoft's BAA
Microsoft includes a HIPAA BAA in its Product Terms for in-scope services. Foundry serves GPT, Claude, Grok and DeepSeek, but only inside Azure, and third-party Foundry models can carry different terms. No Gemini.
Best for: teams fully committed to Azure.
5. Bifrost (Maxim AI): open-source gateway, no inference
Bifrost tops the Maxim AI ranking for this search, which Maxim publishes about its own product. Bifrost is open-source software you deploy yourself. It does not sell inference, so it does not hold BAAs with model providers. You bring your own provider keys and negotiate a BAA with every provider you call.
Best for: teams that want to run their own gateway and already have BAAs with each provider.
6. Kong AI Gateway: API gateway plugin, no inference
Kong adds LLM routing to its API gateway. Like Bifrost, it routes to provider accounts you own, so the BAAs are yours to get, one provider at a time.
Best for: teams that already run Kong for their APIs.
What about OpenRouter, Portkey and LiteLLM?
None of them covers inference under a BAA. OpenRouter's enterprise page lists SOC 2 Type II and a DPA, not a BAA. Portkey stores your own provider keys as virtual keys; its Enterprise BAA covers Portkey, not the models. LiteLLM is an open-source proxy you run with your own keys.
Healthcare AI gateway comparison
Gateway
Who signs the BAA for inference
BAA-covered providers
Gemini under BAA
Fees
Concentrate
Concentrate
7
Yes
$499/month with a BAA (no subscription fee without one); no per-request fees or token markup
Aptible
Aptible
3
No
$200/month plus tokens above list rates
Amazon Bedrock
AWS
AWS only
No
AWS rates
Azure AI Foundry
Microsoft
Azure only
No
Azure rates
Bifrost
You, with each provider
Bring your own
Only with your own Google BAA
Self-hosted
Kong AI Gateway
You, with each provider
Bring your own
Only with your own Google BAA
Kong pricing
As of November 2026. See Model Fortress for Concentrate's live BAA-eligible list.
How do I get started with a healthcare AI gateway on Concentrate?
One API in front of many LLMs that keeps PHI on BAA-covered routes, with zero data retention, no prompt storage, access controls and spend limits.
Which healthcare AI gateways sign a BAA that covers inference?
Among the gateways in this list, Concentrate (7 providers) and Aptible (3 providers). Bedrock and Azure cover their own clouds. Bifrost, Kong, Portkey and LiteLLM leave the provider BAAs to you.
Is an AI gateway HIPAA compliant?
HIPAA does not certify gateways. It depends on the deployment: a signed BAA, a HIPAA-enabled organization and routing only to BAA-eligible providers.
Does Concentrate sign a BAA?
Yes. BAA use is $499/month, with no per-request fees and no markup on tokens. Non-BAA use has no subscription fee. Start at concentrate.ai/compliance.
Can I use Gemini with PHI through a healthcare AI gateway?
Yes, on Concentrate through Google Vertex AI under the same BAA, including Gemini 4 Argon and Gemini 3.8 Flash.
Does one BAA cover every model?
It covers every model on a BAA-eligible provider. Check Model Fortress for the live list.
Blog