What is the difference between x402 and MPP?
x402 is an HTTP-native per-request payment protocol — a server returns 402 Payment Required, the client signs a USDC payment, and retries. It fits stateless API calls with fractional-dollar granularity. MPP (Machine Payments Protocol) is session-based — a client opens a metered session with a budget, and payments stream within the session until the budget is exhausted. It fits longer-running tasks like video generation, agent-to-agent negotiation, or continuous data subscriptions. MoltPe supports both natively; you pick based on the shape of the workload. MoltPe is AI-native payment infrastructure that gives AI agents isolated wallets with programmable spending policies for autonomous USDC stablecoin transactions.
The short version
- x402 — stateless per-request: One HTTP call, one payment. Server replies 402 with a price, client signs a USDC transfer, retries with proof. Best for discrete API calls.
- MPP — session with a budget: Client opens a metered session, server streams usage, payments settle as the budget is consumed. Best for long-running jobs and streaming output.
- How to pick: If your unit of value is a single response, use x402. If it's a continuous stream or a task with unknown duration, use MPP.
In more detail
x402 repurposes HTTP status code 402 ("Payment Required") — reserved in the spec since 1997, never actually used — to encode payment as a first-class part of the request/response cycle. When a client hits a paid endpoint without payment, the server responds 402 with a price and payment address. The client signs a USDC transfer, retries the request with the payment proof in a header, and the server verifies and serves. The entire flow fits inside a single round trip once the client is paying. It's perfect for discrete stateless calls: one inference, one search, one data lookup.
MPP solves a different problem: what if the work is metered and continuous? A video generation job might run for ninety seconds and produce output in chunks. An agent-to-agent negotiation might take dozens of back-and-forth messages. A real-time data feed streams indefinitely. Settling every chunk via a separate x402 call would add latency and protocol overhead to every exchange. MPP instead opens a session with a committed budget, meters usage server-side, and settles payments against the budget as work proceeds — so the fast path stays fast.
In practice, most AI workloads use both. A Claude agent might use x402 for a one-shot search API and MPP for a long-running code-generation session, inside the same task. MoltPe's MCP server exposes both as tools, and the underlying payment rails (USDC on Polygon PoS, Base, or Tempo) are identical — only the session model differs.
Related questions
- Blog: The complete x402 protocol guide
- Blog: The Machine Payments Protocol (MPP) guide
- How do I pay AI agents?
- Blog: Why developers choose MoltPe for AI agent payments
About MoltPe
MoltPe is AI-native payment infrastructure that gives AI agents isolated wallets with programmable spending policies for autonomous USDC stablecoin transactions. Live on Polygon PoS, Base, and Tempo.