Usage-based pricing looks simple from the outside. But deciding on this pricing model is just the first step and involves decisions about billable units, tier structures, and how to account for variability. These choices can have huge implications for revenue, flexibility, and customer satisfaction, and the right approach varies across industries, buyer types, and cost structures.

Below, we’ll break down 10 real-world examples of usage-based pricing across the technology industry for AI, application programming interfaces (APIs), infrastructure, and automation.

Highlights

• The billable unit a company chooses—such as tokens, tasks, or gigabyte-seconds (GB-seconds)—shapes the customer experience as much as the price itself.

• API and infrastructure products work well with pure pay-as-you-go (PAYG) models, while automation and data products are more likely to use capped subscriptions or hybrid committed models.

• AI companies tend to use per-token pricing with predictability wrappers, which reward customers who can predict or shape their usage.

What are some pay-as-you-go and usage-based pricing examples in 2026?

Below are real usage-based and PAYG pricing models used by companies offering software-as-a-service (SaaS), AI tools, APIs, automation, and infrastructure. Each case study explains what the product charges for, how it handles variability, and how the pricing appears to customers.

OpenAI API

• Billable unit: Billed by the token. Input and output tokens are priced separately, with different rates per model tier (e.g., GPT-4o, o1, o3-mini).

• Packaging: PAYG, with no minimum spend. Batch API calls get a discount. Enterprise customers can negotiate committed-use arrangements. Rate limits create an indirect tier structure.

• Customer experience: A researcher who sends short prompts pays almost nothing. A production app that processes long documents at scale might see costs compound quickly. The separation of input and output tokens provides a lever for fine-tuning spend. Payments are powered by Stripe.

Anthropic Claude API

• Billable unit: Billed by the million token batch, with separate input and output rates. Prompt caching (i.e., storing repeated context) is priced lower, which rewards customers who structure their calls to reuse existing context.

• Packaging: PAYG at the API level, with no minimum. Claude’s customer plans layer flat-rate subscriptions on top for users who aren’t developers.

• Customer experience: Thanks to the prompt caching mechanic, customers who carefully design their prompts can save money. Payments are powered by Stripe.

Twilio

• Billable unit: Can be billed per message sent or received, per minute of voice, or per phone number provisioned. Each communication type has its own rate card.

• Packaging: PAYG, with no monthly minimum, and volume discounts that apply automatically at defined thresholds. Committed-use volume pricing is available for high-spend accounts.

• Customer experience: Cost scales predictably. A startup that sends 500 text message (SMS) verifications a month pays much less than a logistics company that sends millions of delivery alerts per week. But tracking overall spend can be complicated due to the multidimensional rate card, which varies price by message type, destination country, and carrier. Payments are powered by Stripe.

Datadog

• Billable unit: Each product line has its own billable unit, such as hosts monitored per hour, custom metrics ingested, log gigabytes indexed, or application performance monitoring (APM) traces analyzed.

• Packaging: Hybrid, with most midmarket and enterprise customers choosing annual committed contracts. On-demand rates are available but steep relative to committed pricing. Many product bundles are available (e.g., “Infrastructure + APM + Logs”).

• Customer experience: Customers typically commit to a host count, then pay for any overages. The multiproduct structure means that a single customer can have five or six separate usage meters running simultaneously. Payments are powered by Stripe.

Amazon Web Services (AWS) Lambda

• Billable unit: Billed by the number of requests plus duration, measured in GB-seconds (memory allocated multiplied by execution time).

• Packaging: A permanent free tier covers a customer’s first million requests and 400,000 GB-seconds per month. Beyond that, it’s PAYG with no minimum or subscription.

• Customer experience: Customers pay only when something runs so cost matches usage. A high-traffic API endpoint can generate more charges during peak hours and nearly zero charges overnight. Refining memory allocation reduces costs.

Snowflake

• Billable unit: Billed by the credit (a computing abstraction that maps to virtual warehouse size and runtime). Storage is priced separately, per terabyte per month.

• Packaging: PAYG with no minimum. Customers typically prepurchase credits at a discount (annual or multiyear). Warehouses pause when idle thanks to automatic suspension features, which makes usage variable.

• Customer experience: Two customers with identical data volumes can have significantly different bills depending on how they schedule queries and size their warehouses. Customers must understand the credit abstraction to meaningfully interpret spend. Payments are powered by Stripe.

Zapier

• Billable unit: Billed by the task. Each action a “Zap” completes counts as one task.

• Packaging: Tiered subscription plans, with monthly task limits. Overage either pauses automation or activates an upgrade prompt. A free tier allows limited tasks.

• Customer experience: Because customers buy a task allowance instead of paying a per-task rate, heavy users can burn through their monthly allowances quicker than expected.

Replicate

• Billable unit: Billed per second of computing time. The per-second rates depend on the specific hardware tier (e.g., CPU, T4 GPU, A100) used by the model.

• Packaging: PAYG with no subscription or minimum. Users can cache models to reduce cold-start latency, which shortens the wait time but doesn’t change the billing unit.

• Customer experience: The hardware tier pricing gives developers a concrete way to think about cost, because they’re essentially renting a specific machine. Payments are powered by Stripe.

Segment

• Billable unit: Billed according to monthly tracked users (MTUs), unique users whose events flow through the platform in a given month.

• Packaging: Tiered plans based on MTU bands. The free tier caps at 1,000 MTUs. Paid plans scale by user volume, while large deployments use custom enterprise contracts. Annual billing is standard at scale.

• Customer experience: The MTU model means a B2C app with millions of casual users (many of whom generate only a handful of events) can reach a higher MTU count than a B2B app with more total events but fewer distinct users. The unit definition creates some counterintuitive outcomes depending on customer type.

Make

• Billable unit: Billed by the operation. Each module execution within a scenario counts as one operation.

• Packaging: Customers pay for tiered monthly plans with operation limits. Unused operations don’t roll over.

• Customer experience: Make’s per-operation pricing makes cost modeling more involved. A developer who builds complex, multibranch automations must predict the overall operation count from the beginning to estimate costs.

How do billable units and packaging strategies compare across industries?

Below is a grid that breaks down the above examples for easier comparison. It lists the primary buyer, billable unit, and packaging of each product, as well as its predictability strategies.

What do these examples reveal about who chooses pay-as-you-go?

A few patterns emerge across the examples above. Products within the same industry tend to have similar methods for usage-based pricing.

API and infrastructure products work well with PAYG

OpenAI, Anthropic, Twilio, AWS Lambda, and Replicate all use PAYG as their default. Each provides a product that works in clearly countable units (e.g., tokens, requests, seconds), has a customer base that varies widely in its usage needs, and is comfortable with variable costs. When the billable unit is easy to define and usage is genuinely unpredictable, PAYG is the natural fit.

Variable automation workload products are split between PAYG and capped subscriptions

Zapier and Make both charge for executions but sell monthly allowances instead of using pure PAYG. This likely reflects their buyer profiles. Operations teams and nontechnical users prefer a predictable monthly line item, even if it means some waste.

Data and monitoring products gravitate towards hybrid models

Datadog and Snowflake both allow variable usage but steer customers towards committed arrangements. These products embed deeply into infrastructure, and usage tends to grow over time rather than peak unpredictably. Customers also benefit from the planning exercise that a committed contract requires.

What are some usage-based pricing patterns across different AI companies?

AI products have largely converged on a recognizable set of packaging patterns. In our examples, they show up clearly across OpenAI, Anthropic, and Replicate.

Per-token pricing is the dominant unit for language models

Language models usually price input and output tokens separately. This reflects a real cost difference on the provider side and gives buyers a concrete optimization target. The split also means that technically sophisticated customers can often reduce costs just by restructuring prompts.

Hardware tiers create a second pricing dimension for inference

With Replicate, the hardware a customer uses explicitly determines their per-second rate. OpenAI and Anthropic express the same idea differently: they use model selection (e.g., GPT-4o vs. o3-mini, Claude 3 Opus vs. Claude 3 Haiku) to balance computing intensity.

Predictability wrappers are appearing around usage models

Anthropic’s prompt caching, OpenAI’s batch API discounts, and enterprise commitment structures at both companies all add cost predictability without shifting to a flat subscription billing model. Usage-based pricing stays in place, but customers who can predict or shape their usage are rewarded for doing so.

Premium model tiers resemble feature tiers in traditional SaaS

With conventional SaaS, you pay more for advanced features. With AI API pricing, you pay more for a more capable (or faster) model. This provides a natural upgrade path that doesn’t require separate product packaging.

The fastest way to understand usage-based pricing is to study different examples. The products mentioned in this article illustrate the vast range of possible strategies. The same underlying concept produces per-token rates, credit systems, MTU bands, operation allowances, and GB-second calculations. It all depends on what the product does and who buys it.

At the same time, the unit a company picks (and how it wraps that unit in tiers, caps, and commitments) can tell you a lot about its customer base and cost structure. When you study enough examples, you start to see the logic that underlies even the most unusual rate cards.

How Stripe Billing can help

Stripe Billing lets you bill and manage customers however you want—from simple recurring billing to usage-based billing and sales-negotiated contracts. Start accepting recurring payments globally in minutes—no code required—or build a custom integration using the API.

Stripe Billing can help you:

• Offer flexible pricing: Respond to user demand faster with flexible pricing models, including usage-based, tiered, flat-fee plus overage, and more. Support for coupons, free trials, prorations, and add-ons is built in.

• Expand globally: Increase conversion by offering customers’ preferred payment methods. Stripe supports 100+ local payment methods and 130+ currencies.

• Increase revenue and reduce churn: Improve revenue capture and reduce involuntary churn with Smart Retries and recovery workflow automations. Stripe recovery tools helped users recover over $6.5 billion in revenue in 2024.

• Boost efficiency: Use Stripe’s modular tax, revenue reporting, and data tools to consolidate multiple revenue systems into one. Easily integrate with third-party software.

Learn more about Stripe Billing, or get started today.