> ## Documentation Index
> Fetch the complete documentation index at: https://docs.dgrid.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# x402 API

> DGrid's x402 API lets developers pay per inference call using the x402 micropayment protocol — no API keys or accounts required.

**DGrid x402 API** is DGrid's pay-per-inference access mode for api payments. Instead of creating an account, pre-funding a balance, or managing separate billing flows, developers and autonomous agents can pay LLM inference costs directly through the **x402 payment protocol** as part of the API request lifecycle.

This makes x402 especially useful for:

* **Agents and autonomous workflows** that need to call an LLM and settle payment immediately
* **Server-side tools and scripts** that should avoid account provisioning and manual top-ups
* **On-demand integrations** where the simplest path is "request inference, authorize payment, receive the result"

## Why x402 API Matters

Traditional AI billing usually requires:

* Creating an account with the provider
* Funding a prepaid balance or attaching a billing method
* Managing API keys, quotas, and billing dashboards

With x402 on DGrid, the flow becomes much more direct:

<Tip>
  Request inference → receive payment requirements → sign once → submit payment and inference together
</Tip>

For agentic systems, this is a major simplification. The same runtime that decides to use a model can also authorize and settle the cost of that inference call, without switching into a separate account-management workflow.

## Primary Use Case: Agents Paying for Inference

x402 is designed for the case where an **agent** needs to:

1. Choose a model
2. Call the model API
3. Pay the inference fee
4. Continue execution with the result

This is valuable for:

* AI agents operating in backend workflows
* Web3 agents or protocols that want payment-aware inference
* Multi-step automations that should treat inference as a directly payable resource

## How the Flow Works

x402 on DGrid uses a **two-step request flow**:

1. **Initial request without `x-payment`**
   The server returns `402 Payment Required` plus signable payment requirements.
2. **Client signs and retries**
   The client generates the `x-payment` payload and sends the same inference request again with the payment header.

This pattern keeps payment authorization explicit while still letting the inference call remain simple and automatable.

## Supported Payment Asset

The current x402 endpoint uses:

* **Protocol**: x402
* **Network**: `eip155:56` (BSC)
* **Payment token**: `USD1`
* **Asset contract**: `0x8d0D000Ee44948FC98c9B98A4FA4921476f08B0d`

## Operational Notes

After successful settlement, the service records payment logs into the `x402_payment_logs` table on the backend.

## When to Use x402 vs AI Gateway

Use **AI Gateway** when you want:

* classic API-key access
* unified account-based model routing
* long-running product integrations with standard billing workflows

Use **x402** when you want:

* direct per-request payment
* no pre-funded account balance
* agent-friendly inference plus payment in one flow