In this blog, we will discuss how to choose the right billing metric and how to implement it. Let's first dive into what billing metrics are and how to decide which metrics are best for use with your AI apis and services.

In terms of charging for blockchain API usage, we generally prefer to charge by API call. While this works for many use cases, it's not the best choice for everyone. This is where choosing the right metric and finding the platform to support it comes in. 
What are billing metrics?
In API monetization, including in AI blockchain API monetization, the billing metric is the basic unit of how the value of the API is charged to the customer. This metric is the basis of your pricing model and directly determines how much you charge your customers to use your API. In its simplest form, you can decide to charge users based on the number of ongoing events or API calls. This means measuring the number of API calls a user or company makes to determine how much they can charge. The more common approach to AI apis is to charge based on the token used (input, output, or combination), in which case you would be metered for the token used in each API request.
Here are a few key points to know when trying to understand and decide on billing metrics:
Usage-based: Billing metrics are often related to how customers use apis. This could be the number of blockchain API calls, the amount of data transferred, or the amount of tokens consumed.
Pricing adjustments: API billing metrics should be aligned with their value. For example, if your API provides valuable data insights, your metrics might be related to the number of data records accessed.
Flexibility and granularity: Good billing metrics provide flexibility. Depending on the level of usage, you may have multiple tiers or pricing plans. The granularity of the metrics (for example, per API call versus per thousand calls) allows you to adjust pricing based on a variety of factors.
Example of AI charging indicators
Although each AI platform and API is unique, the way companies use it is often similar. Later, we'll discuss in more detail how to determine which metrics might be most appropriate; However, here are some examples of billing metrics that you might see when it comes to how to charge for your AI API.
By token
This is the most granular billing unit for AI services, especially for large language models (LLM). Each tag represents a piece of text, such as a word or subword. For example, OpenAI's blockchain API accessibility model is priced per 1,000 tokens, with different rates for input and output tokens. This metric is ideal for applications where input/output text lengths vary widely, such as chatbots or content generation. Tracking token usage can help you maintain a tighter coupling between usage and the potential costs of running your platform. The result of going this route is that you and your users can estimate and control costs more accurately.
Call via blockchain API
This metric is smaller than the cost of token-based billing and each API request made to the AI service. For example, some image generation apis charge a fixed price for each image generated, regardless of the complexity of the prompt. This metric is suitable for applications with predictable or fixed-length inputs/outputs, such as image classification or sentiment analysis. Pricing may be easier for non-technical people to understand than token-based pricing, which can be confusing for less technical users. This approach does simplify billing, but may not be as cost-effective for high-volume customers, or if the potential cost of API calls can fluctuate significantly, resulting in potentially lower profits.
Per user
The final billing metric we'll explore is charging based on the number of users accessing the platform, also known as "per-seat pricing." This model charges based on the number of users or "seats" that can access an AI service, usually on a monthly or annual basis. A real-world example is how Notion offers Notion AI plugins for AI features, charging per AI-enabled seat. As you can see from our Notion example, this metric is common in collaborative AI tools or platforms that provide predictable pricing for businesses. The downside of this approach is that it may not be the most cost-effective way to charge infrequent or low-traffic users, resulting in a cost-to-value ratio that can lead to churn or downgrading.
While there are other ways to charge for the use of AI blockchain APIs, these tend to be the most common metrics for charging.