Part 1: How to Design and Operationalize Subscription + Usage Models for AI Applications

What is the best way to monetize AI? That is a core question for both operators, founders and investors these days. The answer to that question has changed over the past few years. Initially most companies focused on pure subscription models but now we are seeing a change to hybrid and usage-based pricing modes. 

In our most recent “Top 20 AI Pricing Pages” breakdown 17/20 companies had a usage-based component in their pricing model. 

And it makes sense. 

Subscription-only models can’t keep up with variable costs, uneven usage, or evolving customer expectations. That’s why more AI products are shifting to hybrid pricing - a mix of subscription + usage.

But how do you design and operationalize it without breaking your product, billing, or customer trust?

We break it down in our new guide.

1. Why Usage-Based Pricing is Essential in AI

Variable costs

AI applications typically incur high and fluctuating infrastructure costs—especially for inference (GPU/TPU time), large-scale storage, and data processing. A usage-based model helps companies pass these costs on in a fair and scalable way, ensuring profitability as usage grows.

Example:

• OpenAI charges per 1,000 tokens processed because each request directly consumes GPU compute.
• Replicate charges based on runtime and hardware used (e.g., GPU class and time per inference).

Align price with customer value

Customers perceive value in different ways—based on how many documents they process, videos they generate, or API calls they make. Usage-based pricing ties the cost to actual output or impact, making it easier to justify spend and increasing customer satisfaction.

Example:

• ElevenLabs charges based on character count for voice generation—users creating more content pay more.
• Runway ML uses credits tied to video length and resolution, aligning pricing with content output.

2. Choosing the Right Usage-Based Metric

Many argue that choosing the right pricing metric is the most important part of getting pricing right. It defines what you charge for and how customers can grow with you. The metric should closely align with the value customers receive—ideally, the more they use, the more value they get.

Value Metrics can be either functional or outcome based. A few examples below:

How do you find the right pricing metric? It typically involves a combination of steps. Start by identifying a few potential value metrics—these often come from analyzing your product and reviewing how industry peers price. Once you’ve narrowed down 5–10 options, validate them with existing and prospective customers to ensure they align with the value users perceive.

Finally, I like to ask these three questions to evaluate each metric:

Examples of usage based value metrics for AI applications

• Credits
• Characters
• Minutes of Video
• Storage (MB)

3. Layering Usage on Top of Subscriptions

Why Hybrid based pricing instead of just subscription/Usage

A pure subscription model can be too rigid, while pure usage-based pricing can feel unpredictable. Hybrid pricing combines the best of both: a base fee that covers fixed costs and guarantees some revenue, plus usage-based components that scale with the customer’s activity and success.

• Hybrid examples:
• Base plan includes X usage → overages charged per unit
• Tiered subscription + usage caps
• Seat-based pricing + usage charges

Designing effective hybrid pricing for AI isn’t just about choosing metrics—it’s about aligning value, cost, and experience. In Part 2, we’ll dive into how to operationalize your strategy: instrumentation, billing systems, and customer communication. Stay tuned!