“Per-request billing” often describes a broader shift from fixed request bundles to charges that respond to measured usage. It does not necessarily mean a flat fee for every API call: providers may meter tokens, sell usage credits, bill against prepaid balances, invoice after use, or combine reserved capacity with pay-as-you-go overages.
How does API pricing work?
An API provider defines what counts as billable usage, measures it, applies the relevant rates and plan rules, and collects payment under a separate settlement arrangement. The meter and the payment method are different parts of the bill.
As an Amazon Associate I earn from qualifying purchases.
- Meter: A provider may count requests, input or output tokens, cached tokens, reserved capacity, or a combination. For AI APIs, one request can consume very different resources from another.
- Rate: The price can vary by model, service tier, token type, modality, cache treatment, or workload.
- Settlement: Customers may pay from a prepaid balance, receive a postpaid invoice, or buy contracted capacity. A prepaid credit balance can still be consumed at variable usage-based rates.
That is why “per-request billing” is an imprecise shorthand. A request may be the transaction that triggers usage, while tokens or another unit determine the charge.
Why move away from bundles?
A fixed request allowance can make unlike workloads look equivalent. A short chat and a long coding-agent session may each count as one request while consuming very different amounts of compute and tokens.
#1 Best Overall
- FOR Small Facility, Complex, Housing, Arcade
- ONE-TIME-PURCHASE; Small Investment
- TOTAL 63 Features (Modules, 22 Reports)
- Unit, Staff; Member Maintenance & Reporting
- Request Trial, Try Features & Decide !
When GitHub announced that Copilot plans would move from premium request units to usage-based billing, it said the new GitHub AI Credits would track input, output, and cached token consumption against published model API rates. The change was announced April 27, 2026, for June 1, 2026; GitHub said base plan prices were not changing in that announcement. GitHub described its rationale as aligning pricing with actual usage and supporting a sustainable, reliable service. That is the provider’s stated reason, not independent proof of the outcome. GitHub’s announcement
Usage-sensitive meters can make workload differences visible in the bill instead of hiding them inside a bundle. They also make the customer’s cost more dependent on how much is sent, generated, cached, or processed.
What “usage-based” can mean in practice
These examples illustrate distinct billing designs, not a universal industry migration. Policies can vary by model, account, geography, and contract, so check the current documentation and terms for the service you use.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →GitHub Copilot: token-linked AI Credits
GitHub’s April 2026 announcement described replacing premium request units with AI Credits charged according to input, output, and cached token consumption at published model API rates. The announcement said base plan prices would remain unchanged, but that does not make every workload a fixed-price request: usage within a plan can still vary. GitHub’s announcement
Google Gemini API: token metering with prepaid or postpaid settlement
Google says Gemini API Prepay and Postpay billing plans began taking effect March 23, 2026. Prepay deducts charges from a credit balance; Postpay accrues usage and charges at month-end or when the account reaches its assigned spend cap. Google documents billing dimensions that include input, output, and cached token counts, as well as cached-token storage duration. Gemini API billing documentation
The payment plan does not by itself tell you the unit price. Google’s rate card lists model- and workload-specific token prices and includes future effective dates; some listed rates change after December 31, 2026. Check the live rate card for the model and modality you plan to use rather than applying one model’s price to all calls. Gemini API pricing
OpenAI Scale Tier: reserved capacity plus overages
OpenAI describes Scale Tier as a hybrid arrangement for eligible enterprise customers and supported models, not a generally available API plan. Customers buy token capacity for one model snapshot for a minimum of 30 days; billing starts when the token units are allocated. Usage above the entitlement is billed at pay-as-you-go rates under the interval rules described by OpenAI. OpenAI Scale Tier documentation
Anthropic: credits or monthly invoicing
Anthropic’s API billing help describes prepaid usage credits and says organizations with an invoicing arrangement are billed monthly instead. This is an example of how settlement can differ while usage is still metered: credits describe how payment is funded, not necessarily a flat bundle of API calls. Anthropic API billing help
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What to compare before choosing an API plan
Compare the provider’s rate card and contract against your actual workload. “Credits” do not mean the same thing across providers, and a headline request allowance may not capture the cost of a long or context-heavy task.
| Billing detail | What to check |
|---|---|
| Meter | Whether charges count requests, input or output tokens, reserved capacity, or a combination. |
| Token and modality treatment | Separate rates for input, output, cached tokens, cache storage, images, audio, video, and tool use where applicable. |
| Settlement timing | Whether usage draws down a prepaid balance, triggers an auto-reload, is invoiced after use, or is covered by a contract commitment. |
| Limits | Request and token rate limits, quota tiers, spend caps, and what happens when a limit or balance is exhausted. |
| Overages | Whether work can continue while usage reporting or cap enforcement catches up, and how excess consumption is priced. |
| Predictability | How often usage is reported, whether forecasts are available, and how task length or context size changes the bill. |
| Eligibility and term | Geography, account tier, enterprise qualification, model coverage, commitment length, and credit or capacity expiration. |
How to estimate and control usage-based API costs
- Identify the meter and applicable rate card. Confirm the model, tier, region, modality, and effective date, then distinguish input, output, cached-token, and other billable dimensions.
- Measure representative workloads. Look beyond request counts. Include short and long tasks, context-heavy prompts, expected output length, and any caching or tool use your application relies on.
- Choose settlement and limits deliberately. Check prepaid balance behavior or invoice timing, auto-reload settings, spend caps, quota limits, and what happens when a limit is reached.
- Check visibility and enforcement delays. Find out how frequently usage data updates and whether ongoing or long-running requests can continue consuming resources before a cap is applied.
- Recheck when models or rates change. Prices and effective dates can differ by model and workload; update estimates when your provider changes its rate card or your usage mix.
A spend cap is not automatically a real-time stop switch. If the provider documents reporting or enforcement delays, account for possible usage during that interval rather than treating the displayed balance as instantaneous.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Recommended Free Tools
