Pricing

Pay for consumption.
Not for the wiring.

Connectors, OAuth and workflows are the infrastructure. They are not where Agent200 charges you. You pay for the paid AI models and third-party services your agent actually uses.

One statement You are billed for the AI models and paid services your agent uses, on a single Agent200 account, at the provider's usage price. Connectors, OAuth and workflows add nothing to it.

The pricing model

A request can touch both lanes. Only one of them costs money.

One run() can read a user's authorized data and call a model in the same request. The connector work adds nothing to your bill. The model request is billed at the provider's usage price.

Connector operations

No chargeable provider cost behind the operation

  • Slack Retrieve authorized messages
  • Gmail Read authorized emails

On your bill

No connector fee

You use Agent200's integration infrastructure without paying a separate fee for the connector.

Paid models & services

The underlying provider charges for the request

  • openai/gpt-5.6-sol A model request
  • Tavily A web search

On your bill

Provider-level usage price

Billed through your Agent200 account, rather than with a visible markup added on top.

Providers set their own rules
An external provider may still impose its own commercial requirements, quotas or usage charges. Agent200 cannot waive those.
Not an unlimited-usage promise
The absence of a connector fee is a pricing principle. It does not establish a commitment to unlimited free usage.

One billing relationship

Every provider you add is another account, key and invoice.

An agent that uses two models and a search service usually means three vendor relationships to open, fund and reconcile. Through Agent200 that consumption is managed in one.

  • One key for supported models and services

    Your Agent200 API key is the only credential your app needs for them. OAuth for your users' own accounts still applies.

  • One place for what you consumed

    Agent200 accounts for the paid usage your app generates and charges it through the platform.

  • Visibility in the dashboard

    The dashboard shows the models and services your app consumed, together with the charges that go with them.

Without Agent200

  • Model provider

    Account API key Invoice

  • Search provider

    Account API key Invoice

  • Another model provider

    Account API key Invoice

Agent200 · usage

  • AI models

    usage-based
  • Paid third-party services

    usage-based
  • Connector operations

    No connector fee

One account, one API key, one billing relationship.

Illustrative One statement covers every model and paid service your agent uses through Agent200.

How you pay

Pay for what runs.
The rest comes with the key.

Every paid model and service your agent uses lands on one Agent200 statement. Everything that connects your agent to them is part of the platform.

On your statement,
at the provider's price.

Charged to your Agent200 account, with usage and charges in your dashboard. Your end users are never billed.

  • AI model requests

    Every request your agent sends to a supported model.

    Usage-based

  • Paid services

    Search, data and other third-party services that charge for usage.

    Usage-based

Included with your key,
with no separate fee.

  • Connectors

    Every supported connector, without a connector fee on top.

    Included

  • Managed OAuth

    Callbacks, token storage and refresh for the accounts your users authorize.

    Included

  • The execution loop

    Tool calling inside every run(). Only the model requests it makes are billed.

    Included

  • Custom workflows

    Build a capability once, then call it by name from your code.

    Included

Sizing this up for a project? Tell us which models and services your agent will use and we will walk you through what lands on the statement.

Ask about pricing

FAQ

Questions developers ask first.

Something else? Contact us.

Do I need my own OpenAI or Tavily API keys?

No. Supported AI models and paid services are reached with your Agent200 API key, and Agent200 manages the underlying provider relationships. There are no provider accounts to open and no provider keys to bring.

Do connectors cost money?

No. There is no connector fee. When the operation behind a connector has no chargeable provider cost, such as retrieving messages a user has authorized, it adds nothing to your statement.

The external provider's own terms still apply. Its quotas, commercial requirements and any charges it sets are outside Agent200's control, so no connector fee is not a promise of unlimited usage.

How are paid models and services priced?

At the provider's usage price for what your app consumes, with no visible markup added on top. Agent200 accounts for that usage and bills it through your Agent200 account.

Where do I see what my agent has used?

In the dashboard. It is the central place for the paid consumption that runs through Agent200: the models and services your app used, and the charges that go with them.

Does one API key remove the need for OAuth?

No. The API key authenticates your app with Agent200 and covers access to supported models and paid services. Reaching an end user's private data, such as their messages or email, still requires that user to authorize the connection through OAuth, which Agent200 manages.

Do my end users need an Agent200 account?

No. Developers are Agent200's customers. End users interact with the app you built and, when needed, authorize access to their own external accounts through an OAuth flow managed by Agent200. Paid consumption is billed to the developer's account.

Does Agent200 replace my agent framework?

No. You can build your agent with LangGraph, the OpenAI Agents SDK, another framework or custom code, and integrate Agent200 into it. You keep ownership of the agent and its business logic; Agent200 provides and executes its external capabilities.

Does Agent200 schedule recurring runs?

No. Scheduling stays in your app, so your code decides when a request runs. If a brief should be prepared every Sunday, your code schedules that Agent200 request.

200 OK

One key. One bill.
Start with the agent.

Bring the agent you already have and add the capabilities it needs. You pay for the paid services it consumes, not for the wiring.