TL;DR: AI Credits
AI credits are usage-based tokens used by SaaS and AI companies to charge for compute-intensive actions like text, image, and video generation. They function like arcade tokens, allowing providers to align customer costs with the actual variable expense of AI inference.
Key Takeaways
-
The Model: Credits act as a virtual currency. Customers receive an allocation (via subscription or purchase) and spend them based on the “weight” of an AI task—e.g., a simple text summary might cost 1 credit, while a video could cost 50.
-
Ideal Context: Best for products with unpredictable infrastructure costs or those offering premium AI features alongside a core subscription.
-
The Lifecycle: Credits are allocated (bundled or top-up), consumed (tracked via metering), monitored (via dashboards), and eventually reset or expired based on the billing cycle.
-
Pricing Impact: Common models include hybrid subscriptions (base fee + credits), pure usage-based consumption, or tiered plans with increasing allocations.
Implementation Steps
- Define the Credit Unit: Establish what one credit represents (e.g., one API call or a specific action type) to keep pricing granular yet simple.
-
Meter and Record Usage: Implement real-time Metered Billing Software to capture every credit-consuming event accurately.
-
Surface Balances: Provide customer dashboards with low-balance alerts (typically at 80% consumption) to manage expectations and encourage top-ups.
-
Connect to Finance: Integrate usage data with Subscription Invoicing Software to automate resets, rollovers, and overage charges.
The Bottom Line
AI credits solve the “flat-fee” problem where heavy AI users can erode a provider’s margins. By using a credit-based abstraction, companies can maintain healthy gross margins, offer transparent pricing, and scale revenue in lockstep with the compute resources their customers actually consume.
Are you looking to automate the tracking of these credits within your billing system, or are you currently reconciling these consumption patterns manually?
AI credits are usage-based tokens that SaaS and AI companies use to charge customers for compute-intensive features like image generation, text analysis, and video creation. Rather than charging a flat monthly fee, providers allocate or sell credits that customers spend each time they use an AI-powered action—similar to how arcade tokens work.
This guide covers how AI credits function, the pricing models companies use to package them, and the billing and revenue recognition considerations that finance teams encounter when implementing credit-based pricing.
What are AI credits
What are AI credits and why do they matter for SaaS pricing?
AI credits are usage-based tokens or virtual currency used to pay for AI-powered actions within software applications. Think of them like arcade tokens—you purchase or receive an allocation, then spend them each time you use an AI feature. Companies like Adobe Firefly, Microsoft Copilot, Google AI Studio, and Figma have adopted this model to charge customers based on actual consumption rather than flat monthly fees.
The credit system exists because AI tasks vary dramatically in how much computing power they require. A simple text edit might cost one credit, while generating a video could cost fifty or more. Credits typically renew monthly with subscription plans, though unused credits often do not roll over to the next billing period. When customers run out, they can usually purchase additional top-up packs.
- Text generation: Creating or editing content using large language models
- Image generation: Producing visuals from prompts or editing existing images
- Video generation: More compute-intensive tasks that consume significantly more credits
- Code assistance: AI-powered coding suggestions and completions
Why SaaS and AI companies use AI credits
Why do AI companies charge using credits instead of flat fees?
AI inference costs vary dramatically depending on the task. Generating a simple text response might cost a fraction of a cent in compute resources, while producing a high-resolution video could cost several dollars. Credits allow providers to pass variable costs through to the customers who generate them, rather than averaging costs across all users.
For providers, credits create predictable cost recovery and clear usage visibility. For customers, credits offer pay-for-what-you-use transparency and the flexibility to control spend. Both parties can forecast and budget based on credit consumption patterns rather than guessing at flat-rate allocations.
- Cost alignment: Credits pass variable compute costs through to the customer who consumes them
- Usage transparency: Customers see exactly how much each action costs
- Monetization flexibility: Providers can adjust credit pricing without restructuring subscription tiers
- Consumption forecasting: Both parties can plan and budget based on credit usage patterns
How AI Credits Work
The credit lifecycle follows a straightforward pattern: credits are allocated to a customer account, consumed as AI features are used, tracked through a balance, and eventually reset or expire. Let’s walk through each stage.
Credit allocation
Credits are typically assigned in one of three ways: bundled with a subscription plan, purchased separately as top-up packs, or earned through promotions. Some platforms allocate credits per seat, meaning each licensed user gets their own allocation. Others pool credits across an entire organization or team, allowing flexible sharing.
Credit consumption
Different AI actions consume different credit amounts based on their compute intensity. A simple text edit might cost one credit, while generating a video could cost fifty or more. This tiered consumption model reflects the underlying infrastructure costs that the provider incurs.
| AI Action Type | Typical Credit Cost | Example |
|---|---|---|
| Basic text generation | Low (1-2 credits) | Chat responses, summaries |
| Image generation | Medium (5-15 credits) | Creating visuals from prompts |
| Advanced editing | Medium-High (10-25 credits) | Background removal, object generation |
| Video generation | High (50+ credits) | AI-generated video clips |
Credit balances and rollovers
Customers track remaining credits through account dashboards or in-app indicators. Whether unused credits roll over depends on the provider’s policy—some plans allow rollover while others reset balances monthly. Organizations with pooled credits can share allocations across users or business units, which adds flexibility but requires visibility into who is consuming what.
Credit expiration and resets
Subscription credits typically reset at the start of each billing cycle under a “use-it-or-lose-it” model. Purchased or promotional credit packs often have longer expiration windows, sometimes six months or a year. Understanding expiration timelines helps customers avoid losing value they’ve already paid for.
What happens when credits run out
When credits are depleted, providers handle the situation differently. Some pause AI features until credits reset at the next billing cycle. Others allow customers to purchase top-up packs immediately. A third approach continues service at overage rates, billing for additional usage beyond the allocation. This is a key commercial decision that affects both customer experience and revenue predictability.
AI Credit Pricing Models
AI credits can be packaged within various pricing structures, each suited to different customer needs and business models.
Hybrid subscription and credit pricing
The most common model combines a base subscription fee with included credits. Customers pay a monthly or annual fee that includes a credit allocation, then purchase additional credits if they exceed their allowance. Adobe Creative Cloud and Microsoft 365 Copilot use this approach.
Usage-based consumption pricing
In a pure pay-as-you-go model, customers purchase credits and consume them without a subscription wrapper. This works well for customers with unpredictable or highly variable usage patterns who don’t want to commit to a recurring fee.
| Pricing Model | Credit Source | Best For |
|---|---|---|
| Hybrid subscription | Included + purchasable | Predictable base usage with occasional spikes |
| Usage-based | Purchased on demand | Variable, unpredictable usage patterns |
| Tiered plans | Allocated by tier | Customers who can estimate monthly needs |
| Seat-based pools | Per-user allocation | Teams with shared workloads |
| Prepaid packs | Bulk purchase | High-volume users seeking discounts |
Tiered plans with included credits
Good-better-best tiering allocates more credits to higher-priced plans. Google AI offers Plus, Pro, and Ultra tiers, each with progressively larger monthly credit allocations. Customers self-select into the tier that matches their expected usage.
Seat-based credit pools
Some platforms allocate credits per licensed seat. In enterprise settings, per-seat allocations may be pooled across the organization, giving teams flexibility in how credits are consumed across different users and departments.
Prepaid credit packs and top-ups
One-time purchases of additional credits beyond the subscription allocation often come with longer expiration periods and sometimes volume discounts. Prepaid packs appeal to customers who anticipate heavy usage during specific periods.
When AI Credits Make Sense For Your Product
When is credit-based pricing the right choice for a SaaS company?
Credit-based pricing works best when AI features have meaningfully different infrastructure costs per action. If every AI task costs roughly the same to deliver, a simpler flat-rate model might be easier to implement and explain to customers.
Credits also make sense when you want to monetize premium AI capabilities separately from your core product. You might offer unlimited access to basic features while gating advanced AI behind credit consumption. This approach lets customers self-select into higher usage tiers without forcing everyone onto expensive plans.
- Variable compute costs: Your AI features have meaningfully different infrastructure costs per action
- Premium feature monetization: You want to gate advanced AI capabilities without restricting the core product
- Usage-based customer preference: Your customers expect pay-for-what-you-use pricing
- Cost pass-through: You want to recover unpredictable AI inference costs from the customers generating them
How to Implement AI credits in a SaaS Product
Implementing AI credits requires coordination between product, engineering, and finance teams. The goal is a system that accurately tracks consumption, surfaces balances to customers, and connects to billing infrastructure.
Step 1. Define the credit unit
Establish what one credit represents—a single API call, a compute-second, or a specific action type. The unit definition affects everything downstream, so keep it simple enough for customers to understand while granular enough to reflect actual costs.
Step 2. Meter and record usage events
Implement usage tracking that captures every credit-consuming action with timestamp, customer ID, action type, and credit amount. This data can be streamed in real-time or batched to billing systems depending on your accuracy and latency requirements.
Step 3. Surface credit balances to customers
Provide real-time or near-real-time visibility into remaining credits via dashboards, account pages, or in-app indicators. Include alerts when balances run low—80% consumption is a common threshold for notifications.
Step 4. Connect credits to billing and entitlements
Integrate credit tracking with your billing platform to handle resets, top-ups, overages, and entitlement enforcement. Billing software with usage-based capabilities automates the connection between metering and invoicing, reducing manual reconciliation work.
AI Credits in Enterprise Contracts
Enterprise deals add complexity to credit-based pricing through commitments, pooling, and custom terms.
Annual spend commitments and true-ups
Enterprise customers often commit to a minimum annual credit purchase in exchange for volume discounts. If actual consumption falls short of the commitment, a true-up invoice covers the difference at period end. If consumption exceeds the commitment, overage billing applies at contracted rates.
Pooled credits across business units
Large organizations may pool credits across departments, subsidiaries, or geographic regions. This requires visibility into allocation and consumption by entity, which becomes important for internal chargebacks and cost allocation.
Overages and capacity contracts
Overages occur when usage exceeds committed credits. Capacity contracts let customers reserve a credit volume at a negotiated rate, providing cost predictability for high-volume users. Billing systems that handle capacity contracts automatically calculate true-ups and overage charges without manual intervention.
Tracking AI Credits with Ordway
For companies implementing AI credit pricing, Ordway’s usage-based billing capabilities handle the full lifecycle—from credit metering and drawdowns to rollovers, prepaid balances, and revenue recognition under ASC 606. The platform tracks ARR and MRR for credit-based revenue alongside traditional subscription metrics, giving finance teams a unified view of recurring revenue performance.
Frequently Asked Questions
How are AI credits different from API tokens?
AI credits are a billing and pricing unit representing the cost of AI actions, while API tokens are technical units measuring input and output size for language models. One API call might consume thousands of tokens but cost a single credit. Credits abstract away technical complexity for customers while tokens remain relevant for developers monitoring model usage.
Can AI credits be refunded or transferred between customers?
Most AI credit systems do not allow refunds or transfers because credits represent prepaid compute capacity. Policies vary by provider, so reviewing terms of service before purchase is worthwhile. Some enterprise contracts may negotiate custom refund or transfer provisions.
How do AI credits affect gross margin for SaaS companies?
Credits are typically priced above the underlying compute cost to preserve margin. If credit pricing is too low relative to inference costs, gross margin erodes as AI usage grows. Companies typically target healthy gross margins on credit-based revenue, though the exact percentage varies by the cost structure of underlying AI models.
Are AI credits subject to sales tax?
Tax treatment depends on jurisdiction and whether credits are classified as prepaid services or digital goods. Some jurisdictions tax at the time of purchase, others at consumption. Consulting a tax advisor or using tax automation software helps ensure compliance across different markets.
How do AI credits compare to pure usage-based billing?
Credits bundle usage into a purchasable or allocated unit, while pure usage-based billing charges per action without a credit abstraction. Credits simplify customer understanding—”you have 100 credits remaining”—but add accounting complexity for prepaid balances and breakage. Pure usage billing is simpler to account for but can create unpredictable customer bills.
