Escrow vs Exact

At a Glance

exact

Immediate settlementPayment happens instantly when request is made. Simple, fast, minimal overhead.Best for trusted services and low-value transactions.

escrow

Deferred settlementFunds locked until conditions are met. Authorization separate from capture.Best for high-value, usage-based, or disputed transactions.

Feature Comparison

Feature	exact	escrow
Settlement Timing	Immediate	Deferred (conditional)
Payer Protection	None (payment final)	Refund possible until capture
Receiver Risk	No risk (paid upfront)	Must wait for release
Gas Cost	Single transaction	Two transactions (auth + capture)
Complexity	Minimal (direct transfer)	Higher (operator logic)
Variable Pricing	Not supported	Supported (authorize max, capture actual)
Dispute Resolution	Not possible	Supported via operator
Multi-Request Sessions	Every request needs signature	One auth, multiple captures
Trust Required	High (payment irreversible)	Lower (escrow protects payer)

When to Use Each

Use `exact` when:

Low-value transactions

For small purchases where the cost of a dispute exceeds the transaction value.Example: $0.01 API call - not worth the escrow overhead

Trusted relationships

When you trust the service provider and don’t need refund protection.Example: Paying your own infrastructure or established services

Immediate delivery

Service is delivered instantly and verifiable on the spot.Example: Static content, database lookups, instant responses

Minimizing gas costs

Single transaction is cheaper than two (auth + capture).Example: High-frequency micro-transactions where gas matters

Use `escrow` when:

High-value transactions

Significant amounts where you need recourse if service fails.Example: $500 training job - you want protection if it fails

Unknown final cost

Usage-based billing where you don’t know the exact amount upfront.Example: LLM API calls charged by tokens (authorize

10, use

6.50, refund $3.50)

Long-running work

Tasks that take hours or days to complete.Example: Video rendering, data processing, training jobs

Dispute-prone services

Services where quality is subjective or verification is needed.Example: Freelance work, custom deliverables, SLA-based services

Multi-request sessions

Many small requests under one authorization.Example: 1,000 API calls at

0.01 each = one

10 auth, periodic captures

Cost Analysis

exact Scheme

Gas Cost: 1 transaction

ERC-20 transfer: ~50k gas
Total: ~50,000 gas

Example (Base, 0.001 gwei gas price):

Gas cost: ~$0.00005
For $10 payment: 0.0005% overhead

escrow Scheme

Gas Cost: 2 transactions

authorize(): ~150k gas
capture(): ~80k gas
Total: ~230,000 gas

Example (Base, 0.001 gwei gas price):

Gas cost: ~$0.00023
For $10 payment: 0.0023% overhead

Amortization: For multi-request sessions, the auth cost is amortized across many captures. 1 auth + 10 captures is cheaper than 10 exact payments.

Security Comparison

exact Scheme

Risk	Severity	Mitigation
Service non-delivery	High	None - payment is final
Overcharging	Medium	Verify amount before signing
Malicious server	High	Trust required

Trust Model: You must fully trust the service provider. Once payment is sent, you have no recourse.

escrow Scheme

Risk	Severity	Mitigation
Service non-delivery	Low	Refund before capture
Overcharging	None	`maxAmount` enforced on-chain
Malicious operator	Medium	Choose trusted operators, set expiry
Operator disappeared	Low	Payer can reclaim after `authorizationExpiry`

Trust Model: You must trust the operator contract logic, but funds are protected by on-chain invariants.

Operator Selection Critical: The operator contract controls when funds are released. A malicious or buggy operator can lock funds. Always audit operator code or use verified implementations.

User Experience

exact Scheme Flow

Server: "Pay $10"
Client: [Signs payment]
Client: [Sends request with signature]
Server: [Receives payment + delivers service]
Done

Steps: 3 (request → sign → deliver) Latency: Single round-trip

escrow Scheme Flow

Server: "Authorize up to $10"
Client: [Signs authorization]
Client: [Sends request with signature]
Server: [Locks $10 in escrow + delivers service]
Server: [Calls operator to capture actual amount]
Done

Steps: 4 (request → sign → deliver → capture) Latency: Service delivery same as exact, but capture happens async

Perception: From the user’s perspective, escrow and exact feel the same during the request. The capture happens in the background.

Example Scenarios

Scenario 1: Simple API Call

Service: Weather API lookup Cost: $0.01 Trust: High (established provider) Decision: ✅ Use exact Why: Low value, instant delivery, trusted service. Escrow overhead not worth it.

Scenario 2: LLM Agent Session

Service: 100 GPT-4 calls Cost:

0.05-

0.20 per call (variable) Trust: Medium (new provider) Decision: ✅ Use escrow Why: Variable pricing, multiple requests, moderate trust. Authorize $20 max, capture actual usage.

Scenario 3: Training Job

Service: Train ML model on GPU cluster Cost: $500 Duration: 48 hours Decision: ✅ Use escrow Why: High value, long-running, need verification before payment. Release after job completes successfully.

Scenario 4: Freelance Work

Service: Custom logo design Cost: $200 Dispute Risk: High (subjective quality) Decision: ✅ Use escrow with arbiter Why: Subjective deliverable, need dispute resolution. Arbiter operator releases on approval or mediates disputes.

Scenario 5: Micro-Transaction Spam

Service: Rate-limited API (1000 req/sec) Cost: $0.0001 per request Volume: 100,000 requests/day Decision: ✅ Use escrow with batch captures Why: Too many requests to sign individually. Authorize $10 daily, server batches captures hourly.

Performance Comparison

Throughput

Metric	exact	escrow
Requests/second	1000+	1000+ (auth), 100+ (capture)
On-chain TPS impact	High (every request)	Low (periodic captures)
Signature overhead	Per-request	Per-session

Latency

Phase	exact	escrow
Request latency	+50ms (sign + verify)	+50ms (sign + verify)
Settlement latency	Immediate	Async (seconds to days)
Finality	Instant	After capture

Hybrid Approach

You can support both schemes and let clients choose:

{
  "x402Version": 2,
  "accepts": [
    {
      "scheme": "exact",
      "network": "eip155:8453",
      "amount": "10000000",
      "asset": "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913",
      "payTo": "0xReceiver..."
    },
    {
      "scheme": "escrow",
      "network": "eip155:8453",
      "maxAmountRequired": "10000000",
      "asset": "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913",
      "payTo": "0xReceiver...",
      "extra": {
        "escrowAddress": "0xEscrow...",
        "operatorAddress": "0xOperator..."
      }
    }
  ]
}

Client decides based on:

Transaction value
Trust level
Gas price
Urgency

Migration Strategy

From exact to escrow

Existing exact integrations can add escrow support:

Deploy operator contract
Add escrow to accepts array
Keep exact as fallback
Clients upgrade when ready

From escrow to exact

If escrow proves unnecessary:

Add exact to accepts array
Monitor which scheme clients prefer
Deprecate escrow if unused

Start with exact, add escrow when needed. Don’t over-engineer. Most simple services work fine with exact.

Getting Started

Smart Contracts

Roadmap

At a Glance

exact

escrow

Feature Comparison

When to Use Each

Use `exact` when:

Use `escrow` when:

Cost Analysis

exact Scheme

escrow Scheme

Security Comparison

exact Scheme

escrow Scheme

User Experience

exact Scheme Flow

escrow Scheme Flow

Example Scenarios

Scenario 1: Simple API Call

Scenario 2: LLM Agent Session

Scenario 3: Training Job

Scenario 4: Freelance Work

Scenario 5: Micro-Transaction Spam

Performance Comparison

Throughput

Latency

Hybrid Approach

Migration Strategy

From exact to escrow

From escrow to exact

Decision Tree

Next Steps

Overview

Escrow Spec

Smart Contracts

SDK Installation

Getting Started

Smart Contracts

Roadmap

​At a Glance

exact

escrow

​Feature Comparison

​When to Use Each

​Use exact when:

​Use escrow when:

​Cost Analysis

​exact Scheme

​escrow Scheme

​Security Comparison

​exact Scheme

​escrow Scheme

​User Experience

​exact Scheme Flow

​escrow Scheme Flow

​Example Scenarios

​Scenario 1: Simple API Call

​Scenario 2: LLM Agent Session

​Scenario 3: Training Job

​Scenario 4: Freelance Work

​Scenario 5: Micro-Transaction Spam

​Performance Comparison

​Throughput

​Latency

​Hybrid Approach

​Migration Strategy

​From exact to escrow

​From escrow to exact

​Decision Tree

​Next Steps

Overview

Escrow Spec

Smart Contracts

SDK Installation

At a Glance

Feature Comparison

When to Use Each

Use `exact` when:

Use `escrow` when:

Cost Analysis

exact Scheme

escrow Scheme

Security Comparison

exact Scheme

escrow Scheme

User Experience

exact Scheme Flow

escrow Scheme Flow

Example Scenarios

Scenario 1: Simple API Call

Scenario 2: LLM Agent Session

Scenario 3: Training Job

Scenario 4: Freelance Work

Scenario 5: Micro-Transaction Spam

Performance Comparison

Throughput

Latency

Hybrid Approach

Migration Strategy

From exact to escrow

From escrow to exact

Decision Tree

Next Steps