Cadenza Labs · Live Demo
Three interactive receipts. Each one runs live in your browser and returns a real Kaspa mainnet transaction proof. Click a card to see the settlement, the covenant, and the economics.
Cadenza is the only rail that runs x402, MPP, and AP2 — the three live agentic payment protocols — through the same OUSD-KRC20 covenant suite on Kaspa L1. One SDK, one settlement path, three protocol surfaces. Tempo runs one protocol (MPP). Base runs one protocol (x402). Cadenza runs all three.
An AI agent making 10,000,000 LLM inference calls at $0.001/token = $10,000 in real economic spend. Here is what settling that spend actually costs, by rail:
| Rail | Per-tx fee | Transactions | Total fees | Real-time? |
|---|---|---|---|---|
| Stripe (per call) | $0.30 + 2.9% | 10,000,000 | $3,300,000 | yes |
| Ethereum mainnet | ~$5 | 10,000,000 | $50,000,000 | yes |
| USDT on Tron | ~$1 | 10,000,000 | $10,000,000 | yes |
| USDC on Base | ~$0.05 | 10,000,000 | $500,000 | yes |
| Tempo (Paradigm + Stripe L1 · EVM per-tx) | ~$0.005 (sub-cent) | 10,000,000 | $50,000 | yes |
| Stripe (monthly batch) | $0.30 + 2.9% | 1 | $293 | no |
| Cadenza Stream (batched · Kaspa UTXO) | sub-cent / batch | ~100 | $0.50 | yes |
100,000× cheaper than Tempo — the Stripe + Paradigm EVM L1, our closest architectural competitor. 6,600,000× cheaper than Stripe. 100,000,000× cheaper than Ethereum. All in real time, on Kaspa L1. Tempo's per-tx cost is competitive; its EVM state model isn't. Batched settlement across 10M events on a UTXO chain with covenant-enforced state is what closes the ratio.
Each button runs a live receipt through a Cloudflare Worker mimicking the Cadenza SDK. Settlement hashes
for demos 1 and 2 are real Kaspa mainnet transactions — click them to open the explorer.
The flows shown here use AP2 mandates for narrative clarity. The same rail — same SDK, same Stream
engine, same covenant enforcement — serves x402 (HTTP-402 flows) and MPP (Machine Payments Protocol
sessions) through the same code path. Protocol choice is a partner decision, not a Cadenza decision.