Cloudflare Launches Monetization Gateway Beta for Usage-Based AI Agent Payments
Cloudflare has announced the closed beta of its Monetization Gateway, a capability designed to enable businesses to charge AI agents for accessing digital resources based on actual consumption. As AI agents increasingly browse websites, call APIs, access tools and retrieve data on behalf of users, conventional account and subscription-based payment models can create additional steps for both businesses and autonomous systems.
Cloudflare Monetization Gateway is designed to support an agent-based payment model in which an AI agent can discover the price of a digital resource, make a payment and access that resource within the HTTP request flow. The capability allows businesses to apply usage-based pricing to resources such as websites, APIs, Model Context Protocol (MCP) tools and datasets without developing their own payment infrastructure.
Using HTTP 402 for agent payments
The growth of AI agents is creating new requirements for how digital services are accessed and paid for. An agent may need to purchase individual resources while completing a task, such as retrieving market data, conducting a search, calling an API or accessing AI inference.
Monetization Gateway uses the HTTP 402 Payment Required status code to return payment instructions as part of an HTTP request. This removes the need to direct an agent through a conventional checkout process. Once a payment has been authorised and settled, the requested resource can be accessed.
Cloudflare said the Gateway handles payment verification and settlement, payment failures and retries, analytics and changes to the x402 protocol. Payments currently settle on the Base blockchain using USDC, a stablecoin pegged to the US dollar.
Applications across search, AI and financial data
Cloudflare is already using Monetization Gateway across several applications.
- Coudflare AI Gateway: Customers in the US can use x402 to pay for inference at request time for a selection of AI models, providing an alternative to maintaining a prepaid credit balance. AI Gateway can also provide pricing information through its existing pricing engine, including support for more complex pricing controlled by the resource provider.
- Ceramic.ai: The web search provider is using Monetization Gateway to enable AI agents to pay for individual search requests without requiring an API key. Ceramic.ai operates an index of more than 40 billion pages and is designed to return search results to machine callers with low latency.
- Stocktwits: The financial social network is using the Gateway to make market signals available to AI agents on a per-request basis. These include market sentiment, message volume, follower information and trending tickers.
- API2PDF: The PDF-generation API is using Monetization Gateway to provide access without requiring users or AI agents to create an account and obtain an API key. Its variable pricing model allows an agent to receive a maximum price for a request and pay according to actual consumption.
Moving towards machine-based transactions
For businesses, the model provides a way to make individual digital resources available to machine customers while retaining control over pricing and access. Pricing rules can be based on elements of a request, including URLs, headers and query parameters. This allows businesses to apply different pricing structures depending on the resource being accessed.
The approach also reduces the need for an AI agent to establish an account or commercial relationship with every service it uses. Instead, payment can be incorporated directly into the process of requesting access to a resource.
Developing infrastructure for agentic commerce
The launch reflects a broader shift towards AI agents performing tasks with greater autonomy. As agents increasingly interact with external services, they may need to acquire resources such as data, search results, APIs, AI inference and specialised tools without direct human intervention.
Cloudflare plans to expand Monetization Gateway with additional payment rails, transaction logs, agent discovery capabilities and identity-related features. The company said the broader objective is to enable digital transactions to take place as part of the Internet request itself, providing a framework for usage-based interactions between businesses and AI agents.
