core idea
AP2
A verifiable chain of authorization — the user's intent and the agent's mandate to transact.
x402
In-band pay-per-call: price a request over HTTP 402 and pay to proceed.
decision explainer
AP2 vs x402 for agent builders.
Two protocols keep coming up when agents start to transact — and they're routinely treated as rivals when they answer different questions. AP2 is about authorization: proving an agent had a mandate to pay on a user's behalf. x402 is about settlement: paying per call in-band over HTTP 402. a2a can supply the execution layer and signed run evidence behind an integration, but does not natively implement either protocol.
AP2 = authorization · x402 = settlement · complementary
0
core question each
0
you must pick
0
native protocol implementations
the problem
They're framed as a choice. They're really two layers.
Search 'AP2 vs x402' and you'll find them pitched as competing standards for the same slot. That framing sends builders looking for a winner that doesn't exist. AP2 concentrates on the authorization problem — proving a user's intent and the agent's mandate to act. x402 concentrates on the settlement problem — how the payment for a single call actually executes. Treat one as a replacement for the other and you'll under-solve whichever half you skipped.
AP2 answers 'was this agent allowed to pay?' — a delegation and intent question.
x402 answers 'how does this call get paid for?' — an in-band settlement question.
Choosing only one leaves either authorization or settlement unaddressed.
Protocol payment evidence and signed execution evidence are distinct records.
the a2a way
Read them as authorization and settlement — then decide by your risk.
Pick by which problem is hard for you. If provable delegation is the risk, lean AP2-style. If frictionless in-band payment is the need, lean x402-style. a2a can run the call and sign execution facts; protocol-specific authorization and payment proof remain in their integration layers.
AP2: mandates and intentx402: pay-per-call over HTTPDifferent questions, not rivalsWhere a2a can fitAuthorization as scoped grantsExecution evidence stays separate
AP2: mandates and intent
The Agent Payments Protocol (AP2) frames agent commerce around a chain of verifiable authorization — the user's intent and the mandate an agent carries to transact on their behalf. Its center of gravity is provable delegation: proving an agent was authorized to buy.
x402: pay-per-call over HTTP
x402 revives HTTP 402 'Payment Required' so a request can be priced and paid in-band. Its center of gravity is the transaction itself: a machine client pays for a single request and proceeds, no signup and no plan.
Different questions, not rivals
AP2 leans on 'was this agent allowed to pay for this?' x402 leans on 'how does the payment for this call happen?' Many real systems will answer both — authorization and settlement are complementary layers, not competing ones.
Where a2a can fit
a2a cloud can run the underlying agent call under a scoped grant and sign evidence on the API, MCP, and /invoke paths. Direct gateway responses include that evidence; Agent API evidence is persisted. AP2 mandates and the x402 in-band payment handshake require an integration layer.
Authorization as scoped grants
If your concern is AP2-style delegation, a2a's scoped grants provide audience- and TTL-bound authority for delegation. Receipts record the grants that authorized it; they do not claim every local action is grant-mediated.
Execution evidence stays separate
Whichever protocol governs intent or settlement, the execution primitive is the same: an Ed25519-signed receipt binding caller, grant IDs, input hash, result preview, and timing. Settlement amounts live in the adjacent control-plane ledger rather than in the signed receipt payload.
side-by-side
AP2 vs x402, dimension by dimension.
primary question
AP2
Was this agent authorized to make this purchase on the user's behalf?
x402
How does payment for this specific call actually happen?
granularity
AP2
Oriented around a transaction with a mandate behind it — often a considered purchase.
x402
Oriented around a single request — bursty, metered, machine-to-machine calls.
best fit
AP2
Agents shopping or paying on a user's behalf, where provable delegation is the risk.
x402
Agents buying individual API/tool calls, where frictionless per-call settlement is the need.
with a2a
AP2
Map delegation to scoped grants; receipts record the grants that authorized it.
x402
Use an external x402 integration for in-band payment; use a2a's receipt as execution evidence, not proof of payment.
dimension
AP2
x402
core idea
A verifiable chain of authorization — the user's intent and the agent's mandate to transact.
In-band pay-per-call: price a request over HTTP 402 and pay to proceed.
primary question
Was this agent authorized to make this purchase on the user's behalf?
How does payment for this specific call actually happen?
granularity
Oriented around a transaction with a mandate behind it — often a considered purchase.
Oriented around a single request — bursty, metered, machine-to-machine calls.
best fit
Agents shopping or paying on a user's behalf, where provable delegation is the risk.
Agents buying individual API/tool calls, where frictionless per-call settlement is the need.
with a2a
Map delegation to scoped grants; receipts record the grants that authorized it.
Use an external x402 integration for in-band payment; use a2a's receipt as execution evidence, not proof of payment.
questions
Frequently asked.
What's the difference between AP2 and x402?
They answer different questions. AP2 (the Agent Payments Protocol) is about authorization — a verifiable chain proving a user's intent and the mandate an agent carries to transact on their behalf. x402 is about settlement — reviving HTTP 402 so a machine client can pay per call in-band. AP2 asks 'was this allowed?'; x402 asks 'how does the payment happen?'
AP2 vs x402 — which should agent builders use?
It depends on your risk. If your agents shop or pay on a user's behalf and the hard part is proving they were authorized, AP2-style delegation is the fit. If your agents buy individual API or tool calls and the hard part is frictionless per-call settlement, x402-style pay-per-call is the fit. They're complementary, and many systems will use both.
Are AP2 and x402 competitors?
Not really. Authorization and settlement are different layers of the same stack. AP2 concentrates on proving delegation; x402 concentrates on executing the payment for a call. You can authorize an agent with a mandate and still settle the resulting calls per-request — the two approaches address adjacent problems rather than the same one.
How does a2a work with AP2 or x402?
a2a can be the execution layer behind either integration. Runs on the API, MCP, and /invoke entry points sign caller identity or classification, skill, input evidence, the grants that authorized it, outcome or result preview, and timing. Direct gateway responses include evidence; Agent API evidence is persisted. AP2 and x402 are external, and cost remains separate from the signed payload.
Do I have to pick one?
No. Because AP2 and x402 target authorization and settlement respectively, an application can compose them. In an a2a-backed design, the protocol-specific mandate and payment exchanges remain in external integration layers; a2a scoped grants authorize execution and signed receipts provide evidence of what ran.
keep reading
Related guides.
All guides live in the guides index.
don't trust the agent
Keep execution proof separate from payment proof.
An a2a-hosted call can run under a scoped grant and produce Ed25519-signed evidence covering caller identity or classification, skill, input, the grants that authorized it, outcome or result preview, and timing. Native Agent API per-call economics live in a separate ledger row. Add AP2 mandates or x402 through an explicit integration.