Aurora routes AI inference across its own nodes and external networks, signs every result cryptographically, and meters it per client — a working protocol serving real demand, not a slide deck.
Get an API key, top up credits, and call /api/create_task. Pay only for what you use — no key needed to try the free tier.
Turn idle compute into contribution straight from your browser. Earn testnet $AIAU while your device helps power the network.
Launch your web node →Send a prompt through the real queue right now — it’s claimed by a node, answered by a real model, signed, and verified end-to-end.
Run a live inference →create_task → queue → node claims → real model → ECDSA-signed → verified. End-to-end through Aurora.
Redis-Streams task queue with atomic claim and crash-recovery. No PHP, no MySQL polling.
Own inference nodes plus a priority-failover gateway across 12 external AI networks. Provider identities stay private.
Every node signs its output (ECDSA secp256k1); the orchestrator verifies the signature and pins each node to its key.
Per-client accounting, real counters only — no synthetic figures. Token economy is testnet.