Private AI for your next application.
The planned API combines a familiar request format with inference credits and inspectable execution evidence. Explore the planned integration model and follow us for app launch updates.
API, MODELS, AND PROVIDERINTEGRATIONS ARE PLANNED
From credits to a request.
The production API is not available yet. These steps describe the intended integration. App access will open at a later date.
Activate inference credits.
Acquire mCREDIT and activate it into an API balance. Each credit represents $1 of AI usage. Activation burns the credit.
Use an application credential.
The planned API would authenticate requests and meter usage against the balance. Production credentials will become available when developer access opens.
Check the evidence in scope.
Inspect the selected route, its trust assumptions, and any receipt. Attestation concerns specified components; it does not establish factual correctness.
Read the trust modelA familiar API format.
This request demonstrates the planned integration shape. The example.com endpoint is a placeholder, and the model is planned. No real request is made by this page.
curl https://api.example.com/v1/chat/completions \
-H "Authorization: Bearer YOUR_MEMBRA_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "planned-private-model",
"messages": [{
"role": "user",
"content": "Evaluate the hypothesis. Protect the inputs."
}]
}'Know what is available.
The app is in development.
Public app access is not open yet. Follow Membra Network on X for updates on availability. The examples here describe the intended integration.
Integrate protected inference.
Production authentication, model availability, usage pricing, provider routes, and evidence services still need to be defined and implemented.
Build autonomous workflows.
Programmable budgets, policy-bound execution, and protected research workflows are future work. They are not deployed by the current protocol design.