AI that can prove where it ran.
One OpenAI-compatible API in front of Europe's sovereign AI endpoints. Every answer arrives with a signed record of where it was computed and under whose jurisdiction — one you can check yourself.
What NeuralRing is
One API, many sovereign models
If your code speaks OpenAI, it already speaks NeuralRing — change the base URL and key, nothing else.
A rule that cannot be bent
Say “Germany only”, “EU only”, or “no US-owned parents” — and no price or speed option can ever override it.
Proof with every answer
A signed, tamper-evident receipt of where the computation ran. Anyone can check it on a public page.
Where do you fit?
Developers
Ship on sovereign AI without changing your stack.
How you start
- Create a key — free signup, no card.
- Point your OpenAI or Anthropic SDK at https://neuralring.eu/api/v1.
- Add one line — sovereignty: { jurisdiction: ["DE"] } — and every response arrives attested.
Individual users
Chat with Europe's AI models — and see the proof under every answer.
How you start
- Sign up in a minute.
- Open the playground and pick a model — or let the Auto Router pick.
- Click the attestation chip beside any answer to see where it ran — and share the public link with anyone.
Legal, privacy & compliance teams
Turn AI vendor review into a champion: receipts, vendor dossiers, and one policy — in your language, not the API's.
Infrastructure & endpoint providers
You built sovereign capacity. We bring it the developers.
How you start
- Submit your endpoint in the partner portal — it is probed and evidence-checked automatically.
- A human reviews the evidence and approves you into the catalog.
- Earn from federated demand with transparent settlement statements — or run the whole gateway under your own brand.
How it works
One API in front of many sovereign endpoints. You set the rule once — a jurisdiction, a minimum tier, a minimum assurance. Routing, failover and load-balancing all happen inside that rule. If nothing qualifies you get an honest error, never a silent fallback to somewhere you didn't allow.
The rule filters first. Price and speed only ever compete among endpoints that already passed it.
Your app sends a request through a sovereignty-rule gate into the NeuralRing router, which routes only to eligible EU endpoints; an endpoint with a non-EU parent is excluded. The response returns to your app with a signed attestation.
The proof, end to end
Every request writes a record.
Endpoint, jurisdiction, tier, assurance, hops and token counts — metadata only, never the prompt or the answer.
Each record is chained to the one before.
The record embeds a hash of the previous record, and its own hash covers that link — one continuous chain.
Why it can't be quietly altered.
Change one byte anywhere in history and every later fingerprint stops matching. Each record is also Ed25519-signed against a published key, so records can't be re-forged wholesale.
record 1 ✓─record 2 ✓─record 3 ✗ tampered✂record 4 — brokenCheck it yourself, without trusting us.
Three independent ways to verify — none of which require taking our word for it:
- Paste any request ID at verify.neuralring.eu — no account needed.
- Export any date range and verify it offline with the bundled standalone script against the published public key.
- Or click the attestation chip beside any playground answer.
The chain proves the record wasn't altered after the fact; the assurance level states how strongly the underlying claim was substantiated. We never let one imply the other.
Every signal states exactly how substantiated it is:
- self-attested the provider says so
- verified we checked, and kept the evidence
- certified a live third-party certification backs it
Questions, answered honestly
Is this a reseller of OpenAI?
What happens if no endpoint matches my rule?
Do you read my prompts?
What does an attestation actually prove — and not prove?
Can I bring my own provider contract?
sovereign inference, provable by design