Hybrid pricing is a billing model that combines a fixed recurring fee, typically a platform or base subscription, with variable charges that scale based on actual product usage. It gives SaaS and AI companies the revenue predictability of subscriptions alongside the expansion upside of usage-based billing, which is why it has become the fastest-growing pricing model in software: over 60% of SaaS companies now use some form of hybrid pricing, up from under 30% in 2021.

The shift is especially pronounced in AI, where consumption is inherently unpredictable. API calls, token usage, compute minutes, and agent executions don't map to seat counts but customers still want a predictable monthly baseline. Hybrid pricing solves both sides of that equation.

This is the most complete guide to hybrid pricing on the internet. At Flexprice, we build the monetization infrastructure that AI-native companies use to ship hybrid billing in hours, not quarters, and we've distilled everything we know into this page.

TL;DR

• Hybrid pricing = subscription + usage + credits combined into one flexible billing model

• Flat pricing breaks in AI because costs scale with every token, API call, and inference run

• Hybrid gives predictable base revenue while capturing upside as customer usage grows

• It solves the core tradeoff: fairness for customers + predictability for the business

• AI makes per-seat pricing inefficient by reducing headcount but increasing product usage

• Top companies like OpenAI, Vercel, and Cursor already use hybrid as their default model

• Revenue scales automatically with usage without constant plan upgrades or sales touchpoints

• Hybrid improves margins by aligning pricing directly with infrastructure and AI costs

• Core building blocks: base fee, usage metering, credits/tokens, and add-on tiers

• Multiple models exist (subscription + usage, credits, tiered + overage, outcome-based)

• Choosing the right value metric is critical, like tokens, API calls, tasks, or outcomes

• Success depends on a billing infrastructure that supports real-time metering, credits, and flexible pricing changes 

What is hybrid pricing?

Hybrid pricing is a billing model that combines subscription, usage-based, and credit-based pricing into a single plan. It gives businesses the flexibility to charge customers a flat recurring fee for baseline access while billing additionally for actual product or service consumption.

Think of it like your phone bill: you pay a fixed monthly rate for your plan, but when you exceed your limit, you're charged based on how much you use. In SaaS and AI, the same logic applies: a platform fee covers core access, while API calls, token usage, or compute minutes are billed on top.

How it differs from pure subscription and pure usage models

When you hear about hybrid pricing, the one question that comes to your mind is how exactly it differs from pure subscription or usage models. Most pricing models force a trade-off. Hybrid pricing exists because neither extreme really works on its own.

• Flat fee: Predictable revenue, easy to sell, simple to budget. But it breaks when usage varies; heavy users get undercharged, light users feel overcharged.

• Pay-as-you-go: Customers pay for exactly what they use, which feels fair. But revenue becomes volatile, and forecasting gets messy, especially for finance teams.

Hybrid pricing sits in the middle. You get a stable base from subscriptions, while usage-based components scale revenue with actual value delivered.

• Predictability for the business

• Fairness for the customer

• Flexibility as usage grows

It’s not just a compromise, it’s a way to align pricing with how modern products are actually consumed.

The core building blocks of a hybrid model

Hybrid pricing isn’t one fixed structure. It’s a combination of a few core components that you can mix depending on your product. Here are some of the core building blocks that structure the pillar of hybrid pricing:

• Base/platform fee: A fixed recurring charge that runs monthly or annually. It covers access, core features, and sets a predictable revenue floor.

• Usage-based charges: Variable fees tied to consumption, which are API calls, minutes, seats, data processed, etc. This is the place where your revenue scales with value.

• Credits/tokens: Prepaid units that customers can spend over time. Useful for smooth billing, offering discounts upfront, or giving customers more control over spend.

• Add-on tiers: An optional upgrade that includes premium features like higher limits and priority support. It helps you monetize advanced use cases without complicating the base plan.

The real power comes from how you combine these:

• Base fee + usage, which is the most common pricing method used for AI and SaaS

• Credits + overage that is good for predictable yet flexible billing

• Add-ons layered on top that help to capture expansion revenue cleanly

A good hybrid model doesn’t feel complex to the customer. It feels intuitive to pay a base to get started, then scale as you grow.

Why hybrid pricing became the default in 2026

Ask a SaaS founder in 2023 how they priced their product, and the answer was almost always flat-rate subscriptions or per-seat licensing. Simple, predictable, easy to slap on a pricing page. But somewhere between the explosion of AI-native products and the collapse of seat-based economics, hybrid pricing stopped being an experiment and became the only model that actually works.

The market shift: from 27% to 41% in 12 Months

According to Growth Unhinged's 2025 State of B2B Monetization report, hybrid pricing adoption surged from 27% to 41% in just twelve months. During the same window, seat-based pricing dropped from 21% to 15%, and flat-fee subscriptions fell from 29% to 22%. Growth unhinged Chargebee's 2025 report projects hybrid adoption hitting 61% by the end of 2026. Companies that are clinging to single-dimension pricing are paying for it:

• 2.3x higher churn among seat-only models versus hybrid or usage-based billing 

• 40% lower gross margins on AI products priced exclusively per seat 

• 21% median growth rate for hybrid billing companies, outperforming both pure subscription and pure usage-based approaches

If you are still running a single pricing dimension in 2026, you are actively losing customers to competitors who price the way modern buyers expect to buy.

Why AI broke traditional pricing models

The shift to hybrid billing did not happen because VCs published thought pieces. It happened because AI fundamentally changed the cost structure of software, and traditional models could not absorb it.

Every AI feature carries a real, variable infrastructure cost: tokens consumed, GPU cycles burned, inference calls hitting your cloud bill. That creates three failure modes that make flat and per-seat pricing untenable:

• Variable costs with fixed revenue. Flat subscription-based billing means margins erode with every heavy user. One enterprise customer running 50,000 API calls a day could cost more to serve than they pay, and your billing system was never designed to flag that.

• Per-seat pricing punishes AI efficiency. AI promises to replace manual work. If your product reduces headcount, charging per seat cannibalizes itself. Enterprise billing teams see this immediately, and it stalls deals.

• Tiered pricing creates misalignment. One customer uses 200 inference calls a month, another on the same plan uses 200,000. The first feels overcharged, the second destroys your margins. Credit-based billing and usage metering solve this, but only if your system supports hybrid models natively.

This is exactly why IDC forecasts that 70% of software vendors will refactor pricing away from pure per-seat models by 2028.

What top SaaS and AI companies are doing right now

The companies setting the standard are building modular hybrid billing architectures where a base subscription provides predictability and a usage or credit layer captures variable value:

• OpenAI: ChatGPT subscriptions (Free to $200/month) with usage limits per tier, token-based usage-based billing on the API as a separate product, and custom enterprise pricing with unlimited model access and dedicated infrastructure

• Vercel: $20/month base per seat plus metered bandwidth, serverless compute, and edge requests, with a $20 credit buffer smoothing the fixed-to-variable transition

• Cursor: Monthly subscription with credit pools tied to actual model API costs, where premium model selection deducts from credits

• Anthropic: Pro at $20/month and Max up to $200/month for individuals, Team seats from $25/seat/month, and self-serve Enterprise at $20/seat with API-rate usage billed separately

Key benefits of hybrid pricing for SaaS and AI

There is a reason nearly half of all SaaS companies have already made the switch to hybrid pricing. It is not a trend. It is a response to a fundamental shift in how software delivers value, how customers expect to pay for it, and how AI is rewriting the economics of both. Here is what changes when you move beyond a single pricing model.

Revenue scales automatically with customer growth

Think about what happens with flat subscription pricing. A customer signs up at $99 a month. They love your product. They triple their usage, invite their entire team, and push API calls through the roof. And you still collect $99.

Hybrid pricing fixes that by design. The subscription gives you a stable baseline, but the usage layer grows in lockstep with how much value each customer pulls from your product. When they process more documents, run more queries, or trigger more automations, the bill reflects it. Revenue expands without a sales rep picking up the phone, without a renewal negotiation, and without a pricing page redesign.

One of the hardest tradeoffs in pricing is serving a startup on a tight budget and an enterprise with deep pockets using the same product. Flat pricing forces you to pick a lane. The price is too high, and you lose the long tail of small customers who could become your biggest accounts in two years. The price is too low, and you leave enterprise money on the table.

Hybrid pricing removes the tradeoff entirely. The structure works like this:

• For smaller customers: A low base fee (or even a free tier) gets them in the door. They pay for what they use, nothing more. The barrier to trying your product drops to nearly zero.

• For enterprise buyers: Committed spend tiers, volume discounts, and annual contracts give them the predictability their procurement teams demand, while usage components still capture the full scope of their consumption.

Same model, two very different customer experiences. Both feel like the pricing was built for them.

AI features become profitable instead of a cost center

This one matters more in 2026 than it ever has. If your product uses large language models, runs inference on GPUs, or orchestrates AI agents, you know the reality: every API call, every token processed, every agent action costs you real money. Traditional SaaS margins of 80 to 90 percent do not apply here. AI products often operate at 50 to 60 percent gross margins, and that gap can swallow your business if the pricing does not account for it.

Hybrid pricing turns this around. Instead of absorbing variable AI costs inside a flat subscription and hoping the math works out, you meter every token, every compute cycle, every agent task, and bill accordingly. The subscription covers platform access and core features. The usage layer covers the actual cost of delivering AI to each customer. Suddenly, your most active users (the ones consuming the most resources) are also your most profitable accounts, not your most expensive ones.

Reduced churn and higher net revenue retention

Churn often has nothing to do with your product being bad. It happens when customers feel the price does not match the value they receive. Flat pricing creates two types of unhappy customers simultaneously: light users who feel they are overpaying for features they barely touch, and power users whose heavy consumption quietly erodes your margins. In contrast, they feel like they are getting a steal.

Hybrid pricing aligns the bill with reality. Customers who use less pay less, so they never hit that this isn't worth it moment of renewal. Customers who use more pay more, but they also see more value, which makes the higher bill feel earned rather than arbitrary. The result is lower churn across the board and a revenue base that compounds over time instead of leaking.

Types of hybrid pricing models

Not all hybrid models are built the same. The right structure depends on what you sell, who you sell to, and what metric best captures the value your product creates. Below are the five most common configurations, each suited to a different type of business. Most successful companies start with one and evolve over time, so think of these less as rigid categories and more as starting points.

Subscription + usage-based

This is the model most SaaS and AI companies land on first, and for good reason. It is the simplest version of hybrid pricing to understand, sell, and operate.

The customer pays a recurring subscription fee (monthly or annual) for platform access and a set of core features. On top of that, they pay per-unit charges based on actual consumption. The usage metric varies by product: API calls for developer tools, tokens processed for AI platforms, messages sent for communication software, or gigabytes stored for data products.

Why it works: 

The subscription provides revenue predictability for your finance team. The usage component captures the natural expansion as customers scale. It is the closest thing to a "default" hybrid model, and platforms like Flexprice make it straightforward to configure by combining subscription invoicing with real-time event metering on a single plan.

This plan is best for: 

• API-first companies

• Developer tools

• Data platforms

• Any product where consumption naturally varies across customers.

Subscription + credit/token system

Instead of billing per unit after the fact, this model gives customers a monthly credit or token allowance bundled into their subscription. They spend those credits as they use the product. If they run out before the billing cycle ends, they either buy more credits or get charged per-unit for the overage.

This is the model that has taken over the AI industry. OpenAI, Anthropic, and dozens of AI-native startups use it because it gives customers a sense of budget control while still allowing flexible consumption. The psychology matters here: customers feel like they are "spending" from a balance rather than racking up an unpredictable bill.

What makes it tick: A robust wallet and credit system that tracks balances in real time, handles auto top-ups, manages expiration rules, and deducts usage accurately across multiple dimensions. Flexprice, for instance, supports prepaid and promotional credits with configurable conversion rates, so you can define your own unit (like "AI credits") and map it to real currency.

This plan is best for:

• AI platforms

• LLM-based products

• Any product where customers want visibility and cost control.

Tiered subscription + overage charges

This is the "pick your plan" model with a safety valve. Customers choose a tier (Starter, Pro, Enterprise) that comes with a defined set of features and a usage allowance. As long as they stay within their tier's limits, the bill is predictable. If they exceed those limits, overage charges kick in on a per-unit basis.

The beauty of this model is familiarity. Customers already understand tiered pricing from personal experience (think mobile phone plans). Adding overage charges on top means you never have to turn away a customer who outgrows their tier mid-cycle. They simply pay for the extra usage until they are ready to upgrade.

It is best suited for: Products with clear feature differentiation across tiers and predictable usage patterns that occasionally spike.

Platform fee + outcome-based pricing

This is the frontier of hybrid pricing, and it is where things get genuinely interesting for AI companies. Instead of charging for inputs like API calls, tokens, and compute time, you charge for outcomes like tickets resolved, leads generated, documents processed, revenue influenced, and tasks completed.

The customer pays a flat platform fee for access. On top of that, they pay based on measurable business results the product delivers. An AI customer support agent, for example, might cost $200/month for platform access plus $2 per ticket it resolves autonomously. The customer only pays more when the product delivers more tangible value.

Why this is gaining traction: As AI agents become capable of performing entire workflows independently, charging per seat or per API call feels increasingly disconnected from the value being created. Outcome-based pricing ties the bill directly to ROI, which makes it the easiest model to justify to budget holders.

It is best suited for: AI agents, automation platforms, and any product where the output is a measurable business result rather than raw compute.

Freemium + pay-as-you-go

The product-led growth favorite. You offer a genuinely useful free tier with real functionality, but capped usage. Once a customer hits the limit or needs advanced capabilities, usage-based charges activate. There is no upgrade wall that is forcing them into a paid plan. They simply start paying for what they consume beyond the free allowance.

This model works brilliantly for bottom-up adoption, where individual developers or small teams discover the product, fall in love with it, and gradually ramp up usage until the account becomes a meaningful revenue source. The key is making the free tier generous enough to demonstrate value but constrained enough that growing teams naturally cross into paid territory.

The numbers speak for themselves: Companies using freemium with usage-based expansion report significantly higher conversion rates from free to paid compared to traditional free trial models, because the transition feels gradual rather than binary. 

It is best fit for: developer tools, PLG SaaS, collaboration platforms, and any product where individual adoption leads to team-wide expansion.

Hybrid pricing for AI products: what's different

If you have spent any time building a SaaS product in the last decade, you have gotten comfortable with a simple truth: software margins are incredible. The AI has rewritten the economics, and pricing models built for the old world are starting to crack.

Why AI changes everything

Every prompt processed, every inference run, every agent task burns through GPU compute and token processing that shows up on your cloud bill.

When COGS is near zero, you can afford to be lazy about pricing. Charge a flat $99 per month, and even if some customers use ten times more than others, the margin absorbs it. When COGS scales with usage, that flat fee becomes a ticking time bomb. Your heaviest users quietly become the accounts that cost you the most to serve.

Hybrid pricing exists in the AI era because it is the only model that ties revenue to the actual cost of delivery. The subscription covers fixed costs. The usage layer covers variable AI costs. Together, they keep the math honest as you scale.

Pricing AI agents and autonomous systems

AI agents add another layer of complexity. An agent does not just respond to a query. It performs work: researching leads, resolving support tickets, generating reports, writing code, and orchestrating multi-step workflows. In many cases, a single AI agent replaces tasks that previously required a full-time employee.

Charging per seat for a product that eliminates the need for seats is a contradiction that customers notice quickly. Instead, the pricing models emerging for agentic AI fall into three categories:

1. Task-based pricing. The customer pays per completed task. An AI agent that resolves support tickets might cost $1.50 per resolution. An agent that processes invoices might cost $0.50 per document. The customer pays only when the agent delivers a tangible output.
   
   

2. Per-action fees layered on a platform subscription. The customer pays a monthly platform fee for access, configuration, and a base allocation of agent actions. Beyond that allocation, each additional action (API call, workflow execution, document generation) is billed at a per-unit rate.
   
   

3. Outcome-based pricing. The most ambitious model, where the bill is tied to measurable business results. Revenue influenced, time saved, leads qualified, deals closed. This works when the outcomes are clearly attributable, and both sides agree on the measurement.

Token and GPU-based metering

Unlike traditional SaaS, where seats or users are the default metric, AI products generate dozens of potential usage dimensions: input tokens, output tokens, GPU-seconds, model type, context window size, storage consumed, and more.

The key is choosing a metric that is both technically accurate and intuitively understandable to the customer, like:

• Tokens processed work for LLM products because it maps directly to cost.

• API calls work for developer tools

• Documents processed or tasks completed work for vertical AI applications 

Once you have the metric, the requirement is a metering system that ingests events at high volume, aggregates them in real time, and connects seamlessly to billing.

The per-seat paradox in AI

This paradox deserves its own spotlight because it is the single most common trigger for AI companies searching for hybrid pricing.

Imagine a customer service platform that charges $50 per agent seat. The company ships an AI copilot that helps each human agent handle three times more tickets. They reduce their team from 30 agents to 12 and get the same throughput. Their bill drops from $1,500 to $600 per month. The product delivered enormous value, and the vendor's revenue was cut by 60 percent.

Now imagine the same product with hybrid pricing: $200 per month for the platform plus $0.10 per AI-assisted ticket resolution. The customer reduces their team, their seat cost drops, but the AI resolution volume increases. The bill might land at $200 plus $800 in usage, totaling $1,000. The customer saves money compared to the old model. The vendor earns more because pricing reflects value delivered rather than headcount deployed.

This is not a hypothetical scenario. It is happening across every SaaS vertical where AI is automating human workflows. And it is why companies that cling to per-seat pricing in the AI era are watching their best feature become their biggest revenue liability.

How to design your hybrid pricing model

Knowing hybrid pricing is the right approach is one thing. Actually designing it is where most companies stall. The good news is that it follows a repeatable five-step process. You do not need perfection on day one. You need a structure solid enough to launch and then iterate with real data.

1. Identify your value metric

Every hybrid model needs a usage unit, and the right one scales naturally with customer value. Not what is easiest to track internally, but what makes intuitive sense to the person paying.

API calls for API platforms. Documents generated for AI writing tools. Gigabytes processed for data pipelines. Tasks completed for AI agents. A quick test: if a customer uses twice as much, do they get roughly twice the value? If yes, you have a good metric. If the relationship feels loose, keep looking.

2. Set your base fee

The base fee is your guaranteed revenue per account. It should cover platform access, core features, and baseline support.

The strategic question is what to include. A common structure: the base fee covers the product interface, a certain number of user seats, standard integrations, and an included usage allowance, say, 10,000 API calls or 50,000 tokens. Everything beyond falls under usage pricing.

The included allowance matters more than most companies realize. It gives customers a risk-free window before variable charges kick in. Set it too low, and customers feel nickel-and-dimed immediately. Set it too high, and you give away too much value in the base tier.

3. Define usage tiers and rates

Three common approaches to variable pricing: Flat per-unit rate. $0.01 per API call, whether you make 100 or 1,000,000. Simple to understand, simple to bill. Works when the cost per unit is consistent.

Volume-tiered pricing. The rate decreases with scale. First 100,000 calls at $0.01, next 500,000 at $0.008, everything above at $0.005. Rewards growth and incentivizes expansion.

Credit packs. Predefined bundles: $50 for 100,000 calls, $200 for 500,000. Budget certainty for the customer, cash upfront for you. Enterprise buyers prefer volume tiers for the visible scale discount. PLG customers prefer simplicity: flat rates or credit packs.

4. Build for enterprise and SMB separately

The same hybrid structure can serve both segments with different configurations. A startup needs a low base fee, self-serve signup, pay-as-you-go, and no annual commitment. 

An enterprise buyer needs a negotiated annual contract, committed spend minimums with discounts, custom usage rates, and dedicated support baked into the platform fee.

5. Decide on prepaid vs. postpaid

The final structural decision is whether customers pay before or after consumption.

Prepaid, where customers buy a credit balance upfront and draw it down with usage. Full visibility, no bill shock, and you get cash before delivering the service. Dominant among AI-native companies, where usage costs can be unpredictable.

Postpaid, where usage is tracked and charged at period end. More flexible, but creates the risk of unexpected bills.

Many companies offer both. An included credit allowance in the subscription, with postpaid overage charges beyond it. Predictable baseline, expansion revenue captured.

Revenue forecasting under hybrid models

Hybrid pricing changes how your product earns money, and it also changes how your finance team predicts it.

Why traditional MRR forecasting breaks

With flat subscriptions, forecasting is mechanical. You know how many customers you have and what each pays. Growth means new logos, contraction means lost logos.

Hybrid pricing introduces consumption as a variable. When 40 to 60 percent of revenue depends on how much customers use each month, the math gets messy. A customer paying $500 in base fees might generate $200 in usage one month and $2,000 the next. Multiplying last month's total by twelve becomes dangerously misleading.

Metrics that matter: beyond MRR

When you are running a hybrid model, MRR alone does not tell the full story. The metrics that give your board and investors real clarity include:

• Committed ARR tracks only the guaranteed subscription and contract revenue, giving you the floor.

• Usage ARR annualizes the trailing usage revenue, showing the growth trajectory of consumption.

• Net Revenue Retention (NRR) measures how much existing accounts grow or shrink over time, combining churn, contraction, and expansion into a single percentage.

• Expansion Revenue isolates the revenue growth coming from increased usage within existing accounts, separate from new customer acquisition.

These metrics, tracked separately but reported together, paint a picture that a single MRR number simply cannot.

Unit economics and financial modeling

Getting the pricing structure right is half the equation. The other half is understanding whether the numbers actually work.

Calculating CAC and ltv in a hybrid world

In a pure subscription business, LTV is simple: average revenue per account multiplied by average lifespan. Hybrid pricing complicates this because revenue per account grows or occasionally shrinks based on consumption.

The accurate approach is cohort-based LTV modeling. Group customers by signup month, track total revenue subscription plus usage over time, and observe the expansion curve. Most hybrid-priced companies see usage revenue start low during onboarding, then increase steadily as customers integrate deeper. The LTV calculation must capture this ramp, not just the initial payment.

CAC stays the same in principle, but the payback period shifts. Hybrid customers often start with lower payments, so payback looks longer initially. But if usage expansion kicks in by month three or four, effective payback can actually be shorter than flat subscription models with no natural expansion path.

Margin analysis when AI COGS are variable

In traditional SaaS, margins are static, but in AI products, every interaction has a cost. The margin equation becomes:

Gross profit = base fee + usage revenue - fixed costs + variable cost per unit x total usage

More usage means more revenue AND more cost. Profitability depends on the spread. Charge $0.01 per API call with $0.004 inference cost, and your usage margin is 60 percent. Every additional call is profitable. If pricing does not maintain that spread, volume makes you less profitable, not more.

Break-even analysis by customer segment

Not all customers reach profitability at the same speed. A startup on $50/month with light usage might take eight months to cover acquisition costs, but on the other hand, an enterprise on a $5,000 annual commitment might break even in month two.

Model break-even by segment: plan tier, company size, industry, and acquisition channel. This reveals where to concentrate sales spend and which segments need pricing adjustments. Hybrid pricing makes this analysis more actionable than flat pricing because usage data gives you a granular view that subscriptions hide.