New See exactly what you're overpaying AWS in under 60 seconds. Estimate your AWS savings →

Azure Functions Pricing and What Shapes Your Monthly Bill

Azure Functions pricing can start at $0 for on-demand usage within eligible free grants. In the East US, legacy Consumption then costs $0.20 per million executions plus $0.000016/GB-second; Flex on-demand costs $0.40 per million executions plus $0.000026/GB-second. One Premium EP1 instance costs about $157.72 for 730 hours. These USD pay-as-you-go figures exclude storage, monitoring and networking.
Picture of Aman Singh
Technical Content Writer, FinOps

Aman Singh

Technical Content Writer, FinOps
November 20, 2026

·

17 min read
Azure Functions Pricing and What Shapes Your Monthly Bill
In this article
Key takeaways
1
A million executions can cost very different amounts depending on the hosting plan, memory and billable time.
2
Premium and Flex always-ready retain paid capacity between executions. A quiet application can therefore still have a recurring hosting bill.
3
Storage, monitoring and networking add to the hosting subtotal. Savings plans can reduce eligible charges, but they do not discount the whole application bill.
Azure Functions pricing can be hard to predict when an app gets little traffic but still generates charges. Functions runs code in response to events, while its hosting plan determines whether unused capacity continues to cost money.

This guide compares hosting plans, with illustrative idle and bursty monthly bills for Flex on-demand and Premium. It explains how memory, execution time and concurrency affect costs, separates hosting charges from supporting services, and shows where savings plans can help.

Azure Functions pricing at a glance

The Azure Functions pricing below uses East US USD public pay-as-you-go rates checked November 11, 2026 (UTC) through Microsoft’s Retail Prices API. 

The query uses API version 2023-01-01-preview, filtered by serviceName eq ‘Functions’, armRegionName eq ‘eastus’ and priceType eq ‘Consumption’. Monthly estimates assume 730 hours; taxes, currency conversion and negotiated discounts are excluded. 
Hosting option Paid rates Monthly free grant Main consideration
Flex on-demand $0.40/million executions + $0.000026/GB-s 250,000 executions + 100,000 GB-s Scales to zero
Flex always-ready $0.000004/GB-s baseline + $0.000016/GB-s active execution + $0.40/million executions None Retained capacity costs extra
Consumption legacy $0.20/million executions + $0.000016/GB-s 1 million executions + 400,000 GB-s Relevant to existing and Windows-dependent apps
Elastic Premium $0.173/vCPU-hour + $0.0123/GiB-hour None At least one billed instance
Dedicated Underlying App Service plan rates No Functions execution grant Shares provisioned plan capacity
Microsoft’s plan-specific free-grant rules apply across function apps in an eligible paid subscription. They are not fresh allowances for every app or temporary account credits.

Flex Consumption is Microsoft’s recommended serverless option for new Linux apps, offering configurable memory and virtual-network integration. Premium supports ready capacity, multiple apps and custom Linux containers. Check execution limits: Consumption caps run at ten minutes, while HTTP responses have a separate 230-second limit.

Dedicated hosting can use spare capacity in an existing plan, but additional demand may require scaling. Our App Service pricing guide explains that underlying bill; Free and Shared App Service tiers cannot host Functions.

Linux Consumption retires September 30, 2028; Windows Consumption is unaffected. Containerized Functions on Azure Container Apps follow that hosting model’s separate charges.

How Azure Functions billing works

An Azure Functions pricing estimate needs billable resource quantities, not just incoming requests.

Executions memory and billable time

An execution is one function run. A request can trigger several runs, and retries can add work. One GB-second means 1 GB of memory for one billable second; a vCPU is a virtual processor core.

Legacy Consumption uses measured memory, rounded up in 128 MB increments, with 128 MB and 100 ms minima. Flex bills configured instance memory during active periods. Its minimum billable period is one second, then time rounds up in 100 ms increments.

Flex offers 512, 2,048 and 4,096 MB instances. CPU increases with size; there is no separate Flex CPU meter. Ordinary asynchronous waiting remains billable, even when little CPU work happens.

Premium bills allocated cores and memory, without an execution-count fee. EP1 provides one core and 3.5 GB memory; EP2 and EP3 are larger sizes, not App Service P-series plans.

Microsoft labels that capacity in GB; the rate table retains the API’s GiB-hour unit, meaning one gibibyte allocated for one hour.

Concurrency and instance sharing

Concurrency means simultaneous executions sharing an instance. Two one-second runs on a 2 GB Flex instance consume four GB-seconds sequentially, or two if they overlap completely without slowing down. Both still count as two executions.

That illustration excludes overhead. A concurrency setting does not guarantee achieved sharing, and CPU-heavy work may become slower.

Azure Functions pricing examples

These Azure Functions pricing examples are modeled hosting subtotals, not invoices or performance benchmarks. Both tables exclude storage, monitoring, networking, downstream services, taxes and commitment discounts.

Choose a suitable hosting model before comparing costs. Start with Flex on-demand for new Linux apps that can scale to zero. Compare Flex always-ready and Premium when responsiveness requires retained capacity; consider Premium for multiple apps per plan or custom Linux containers. 

Legacy Consumption remains relevant to existing or Windows-dependent apps. Dedicated can suit existing spare App Service capacity or requirements for predictable provisioned capacity. Check operating-system support, networking, execution limits and traffic patterns before estimating. 

Before adapting these examples, record the region, hosting plan, execution count, memory, billable instance time, concurrency, retained capacity and remaining grants. Collect the relevant quantities for the same period:
  • Flex on-demand: On Demand Function Execution Units and On Demand Function Execution Count.
  • Flex always-ready: Always Ready Units, Always Ready Function Execution Units and Always Ready Function Execution Count.
  • Premium: Billed instance-hours by SKU.
  • Supporting services: Storage, telemetry and bandwidth quantities.
Microsoft’s metric-to-meter guidance converts Flex resource-unit metrics from MB-milliseconds to GB-seconds by dividing by 1,024,000. Convert execution counts to millions for the rates shown here.

Idle quiet and bursty bills

Assume queue-processing workloads over 730 hours: Flex uses 2 GB instances, concurrency one and one billable second per execution. No retries or zone redundancy apply. Each scenario independently has its full on-demand grants available.
Scenario and allocation Flex on-demand Premium
Idle, no executions; one EP1 retained $0.00 $157.72
Quiet, 100,000 executions; one EP1 retained $2.60 $157.72
Bursty, ten six-hour windows; five active instances and 1,080,000 executions; Premium retains one EP1 and adds four for 60 hours $53.89 $209.57
For bursty Flex, resource use is 5 × 60 × 3,600 × 2 = 2,160,000 GB-s. After grants, (2,160,000 − 100,000) × $0.000026 + 0.83 × $0.40 = $53.89.

For the quiet case, (100,000 × 2 − 100,000) × $0.000026 = $2.60; the execution count remains within its grant.

Premium EP1 costs $0.173 + 3.5 × $0.0123 = $0.21605/hour. The burst allocation is 730 + 4 × 60 = 970 instance-hours, costing $209.57.

Instance counts are assumptions, not predicted autoscaling. Specifications differ between plans, so the totals do not establish equal performance. Any additional startup, scale-in or billed prewarmed capacity increases the estimate.

What keeping capacity ready costs

Keeping an instance ready adds a baseline charge but lowers its active execution rate. Model one 2 GB instance retained for 730 hours, without overflow. Here, on-demand grants are exhausted; equal execution-count charges are omitted.

Microsoft’s Always Ready Units definition counts retained capacity whether functions execute or not.
Active instance-hours On-demand resources Always-ready resources
100 $18.72 $32.54
300 $56.16 $55.58
600 $112.32 $90.14
Always-ready baseline is 2 × 730 × 3,600 × $0.000004 = $21.024. Add 2 × active hours × 3,600 × $0.000016; on-demand instead uses $0.000026 without a baseline.

Under these assumptions, longer active periods can offset the baseline through the lower execution rate. Available grants or more instances change the comparison; it does not measure responsiveness.

Additional Azure Functions costs

An Azure Functions pricing estimate also needs the services surrounding the code:
  • Storage: Account for required storage and transactions, including workflow state where applicable.
  • Monitoring: Application Insights telemetry has separate ingestion and retention rules.
  • Networking: Include applicable outbound/inter-region transfer and separately deployed networking resources.
  • Downstream services: Databases, messaging and specialized orchestration backends have their own charges; Durable Functions billing needs separate treatment.
Check the Azure Functions storage requirements and Durable Functions billing rules when estimating those supporting resources. Our Azure Monitor pricing guide explains monitoring charges in more detail. 

For example, Microsoft Premium Global Network egress from North America includes 100 GB/month free, then $0.087/GB in the next 10 TB tier. With that allowance still available, 500 GB adds (500 − 100) × $0.087 = $34.80. The allowance is not newly available to each app.

Azure Functions cost optimization

The next step is to find which billed quantities can change without hurting the application:
  1. Check actual meters. Compare Cost Management charges with execution counts and resource-unit metrics. Investigate unexplained allocation before extrapolating a month.
  2. Match the plan to requirements. Consider OS support, networking, execution limits and existing spare capacity before comparing rates.
  3. Test resources and concurrency together. A smaller instance may run longer. Compare cost alongside latency, memory pressure and errors using representative work.
  4. Review retained capacity. Keep always-ready instances where their responsiveness benefit justifies the recurring baseline. Check Premium’s minimum and scale-out allocation too.
  5. Reduce avoidable work and data. Address unnecessary retries and excessive logging, and examine transfer paths. Preserve diagnostic evidence and operational headroom.
Microsoft’s monitoring controls help manage telemetry volume. A lower configured limit alone is not proof of lower cost.

Azure Functions savings plans

Azure Functions pricing includes lower rates through Microsoft’s compute savings plan, a fixed hourly spending commitment over one or three years. Its discounts apply automatically across eligible Azure compute usage within the plan’s scope, rather than being reserved for one Functions app. 

Microsoft lists savings-plan pricing for Flex and Premium. Legacy Consumption has no equivalent savings-plan rates in the checked catalog; Dedicated eligibility depends on its App Service SKU. Storage and network charges remain outside compute savings-plan coverage.

Unused hourly commitment expires rather than rolling into busy hours. Native plans cannot be canceled, exchanged or refunded. Base the commitment on eligible hourly spending you expect to maintain after accounting for existing coverage and planned changes, not a monthly average or peak.

Also read our guide to sizing a Functions Premium savings plan before using your Azure Functions pricing estimate to choose an hourly commitment.

How Usage.ai manages Functions Premium commitments 

After estimating Premium costs, you can assess whether a savings plan fits the usage you expect to maintain. At Usage.ai, our read-only Savings Test reviews Functions Premium usage and existing coverage; purchases require separate authorization.

We purchase and manage supported commitments on your behalf without resizing your application or changing its code.

With our Flex Insured Commitments, teams can access 17% savings on Functions Premium CPU and memory rates through a three-year Azure compute savings plan with Cashback protection for qualifying losses under the applicable agreement.

If an eligible Flex Commitment costs more than equivalent pay-as-you-go usage, we provide Cashback protection for the qualifying difference. Our protection does not cancel the underlying commitment, and we do not automatically protect existing customer-owned commitments as Flex Commitments.

We charge an agreed percentage of realized savings, so the fee applies when verified savings are generated.

Build your Azure Functions estimate

Build your Azure Functions pricing estimate from the selected plan, billable quantities and supporting services. Use the Azure pricing calculator where applicable, then reconcile estimates with actual meters.

If your organization has eligible Premium spend, book a session with our team to arrange a read-only Savings Test.

Usage.ai is a multicloud cost optimization platform for AWS, Microsoft Azure and Google Cloud. Commitments are purchased in customer cloud accounts and remain customer-owned; provider terms and payment obligations continue. Usage.ai manages eligible purchases and renewals within approved controls. Read-only assessment is separate from authorized purchasing, and daily analysis does not require daily purchases. Cashback applies only to qualifying losses on eligible Flex Commitments under the applicable agreement, including eligibility, exclusions, notice, credits or offsets, payout timing and service-exit treatment. Ending managed service does not cancel native commitments or automatically refund unused value. Provider charges are separate from Usage.ai fees.

Azure commitment review
Review your Premium commitment coverage

Arrange a read-only assessment of eligible usage, existing coverage and additional commitment opportunities.

Frequently asked questions

Is Azure Functions free to use?

Azure Functions pricing includes limited monthly free grants for eligible Consumption and Flex on-demand usage. Grants are shared across applicable apps in the subscription. Always-ready, Premium and supporting resources can still generate charges; temporary account credits are a separate benefit.

How much does Azure Functions cost per month?

Calculate Azure Functions pricing from your hosting plan, resource quantities, billable time and remaining allowances. One East US EP1 retained for 730 hours is approximately $157.72 before extras; execution-based plans vary with usage. The examples above show hosting subtotals, not complete application bills.

Does Azure Functions charge when idle?

Scaled-to-zero on-demand instances avoid active compute charges. Flex always-ready retains a baseline charge, Premium keeps at least one billed instance, and Dedicated continues billing the App Service plan. Storage, monitoring and existing commitments may also continue costing money.

Which Azure Functions hosting plan is cheapest?

There is no universal winner. Available grants, memory, billable time, concurrency, required features and retained capacity affect the result. Compare suitable configurations using representative workload assumptions; a lower unit rate or one illustrative total does not establish the best production fit.

How can I reduce Azure Functions costs?

Review memory, runtime, concurrency, ready capacity and supporting services before committing to discounted rates. Test changes against performance requirements, then assess the eligible hourly spending you expect to maintain. Savings plans help only when the commitment is used; they do not remove storage or network charges.

Share
Facebook
X
LinkedIn
Reddit
See your savings first
More from Usage.ai
Latest from our blogs