Residency and retention as hard constraints, not hopes.

Healthcare, finance and public-sector teams inherit org-level data policy in every project. Endpoints that violate region or retention rules are eliminated with a stated reason, and the audit trail records it.

Start with this policy Read the guide

WHAT YOU GET

Org-inherited region and retention ceilings

Zero-retention, no-training endpoints only when the policy says so

Audit entries with route evidence that members cannot edit or delete

MODELS FOR THIS WORKLOAD · LIVE CATALOG

Chat & reasoning Speech-to-text Embeddings

regulated.py · OpenAI SDK, base URL changed

res = client.chat.completions.create(
    model="openai/gpt-5.4",
    messages=msgs,
)

res.yelno   # a dict on the OpenAI SDK
# {
#   "request_id":  "req_…",
#   "provider":    "…",      the vendor that served it
#   "endpoint_id": "ep_…",    region and retention on record
#   "charge": {"nanos": …, "currency": "USD", "state": "final"}
# }

Go live in three steps

Set org data policy

Regions, retention ceiling, and no-training are inherited; projects can only tighten them.

Pin a strict template

eu-strict expands into a versioned policy your compliance team can read: region, retention ceiling, no training.

Export the evidence

Exports list every request's endpoint, region and retention class.

Regulated industries guideproduction in four steps

1. Org data policy

Set permitted regions, a retention ceiling of 0d and no-training at the org level. Projects inherit and can only tighten.

2. Template

Pin eu-strict; export the expanded policy for your compliance team to read.

3. Data policy

Only endpoints inside the region, retention and training rules stay eligible; a request with no eligible endpoint is refused with every elimination named.

4. Audit

Evidence exports list every request's endpoint, region and retention class.

guide copy · the code calls the gateway's OpenAI-compatible routes; model slugs are examples from the catalog · not yet connected