Cloud Run pricing

Cloud Run charges you only for the resources you use, rounded up to the nearest 100 millisecond. Your total Cloud Run bill will be the sum of the resource usage in the pricing table after the free tier is applied.

When setting concurrency higher than one request at a time, multiple requests can share the allocated CPU and memory of an instance.

Outbound internet data transfer uses the Premium Network Service Tier and is charged at Google Cloud networking pricing with a free tier of 1GiB free data transfer within North America per month.

Data transfer to Virtual Private Cloud networks is billed as Data transfer from a VM and charged at Virtual Private Cloud data transfer rates. Serverless VPC Access connectors also charge for the compute required to run them. See Serverless VPC Access pricing.

There is no charge for data transfer to Google Cloud resources in the same region (for example for traffic from one Cloud Run service to another Cloud Run service). There is no charge for data transfer to Media CDN, Cloud CDN and Cloud Load Balancing.

Pricing considerations

When evaluating the pricing of Cloud Run, consider the following:

On-demand and pay per use: Cloud Run provides on-demand capacity and automatically scales instances. Cloud Run does not require pre-provisioning infrastructure to accommodate for anticipated peak usage. Container instances billed by Cloud Run are used container instances.
Total cost of ownership: While Cloud Run charges for compute costs, Cloud Run provides more value. For example, Cloud Run offers zonal redundancy, requires low operations because Site Reliability Engineers do a lot in the background, makes you and your team more productive via its simplicity.
Committed use discounts: The cost of any continuous use of Cloud Run can be lowered by purchasing Committed use discounts. For example, if your Cloud Run service always has one or more active instances, you can lower its cost by committing to at least this amount. Compute flexible committed use discounts apply across GKE, Compute Engine and Cloud Run.

Pricing calculator

You can use the Google Cloud pricing calculator to estimate the cost of using Cloud Run.

Pricing tables

The following pricing tables use the GiB-second unit. A GiB-second means for example running a 1 gibibyte instance for 1 second, or running a 256 mebibyte instance for 4 seconds. The same principle applies for the vCPU-second unit. CUD refers to committed use discounts.

The free tier usage is aggregated across projects by billing account and resets every month; you are billed only for usage past the free tier. The free tier is applied as a spending based discount using Tier 1 pricing.

Cloud Run pricing depends on the selected region. Pricing for Cloud Run services also depends on the CPU allocation configuration.

Pricing for regions in Tier 1

Resources	CPU	Memory	Requests
Services with CPU always allocated Jobs	$0.00001800 / vCPU-second beyond free tier Free: First 240,000 vCPU-seconds free per month FlexCUD⁰: $0.00001296 for 1y FlexCUD⁰: $0.00000972 for 3y	$0.00000200 / GiB-second beyond free tier Free: First 450,000 GiB-seconds free per month FlexCUD⁰: $0.00000144 for 1y FlexCUD⁰: $0.00000108 for 3y	$0
Services with CPU only allocated during request for billed instance time	$0.00002400 / vCPU-second beyond free tier Free: First 180,000 vCPU-seconds free per month CUD¹: $0.00001992 If idle min instance²: $0.00000250	$0.00000250 / GiB-second beyond free tier Free: First 360,000 GiB-seconds free per month CUD¹: $0.000002075 If idle min instance²: $0.00000250	$0.40 / million requests beyond free tier³ Free: 2 million requests free per month CUD¹: $0.332

If you pay in a currency other than USD, the prices listed in your currency on Cloud Platform SKUs apply.

⁰ FlexCUD refers to Compute Flexible Committed Use Discounts.

¹ CUD refers to committed use discounts.

² idle min instance refers to idle billable time for instances kept warm using minimum instances. Idle instances that are not minimum instances are not charged.

³ Requests are only billed when they reach the container after successfully being authenticated, requests denied by IAM policy are not billed.

Pricing for regions in Tier 2

Resources	CPU	Memory	Requests
Services with CPU always allocated Jobs	$0.00002160 / vCPU-second beyond free tier Free: First 200,000 vCPU-seconds free per month FlexCUD⁰: $0.000015552 for 1y FlexCUD⁰: $0.000011664 for 3y	$0.00000240 / GiB-second beyond free tier Free: First 375,000 GiB-seconds free per month FlexCUD⁰: $0.000001728 for 1y FlexCUD⁰: $0.000001296 for 3y	$0
Services with CPU only allocated during request for billed instance time	$0.00003360 / vCPU-second beyond free tier Free: First 128,571 vCPU-seconds free per month CUD¹: $0.000027888 If idle min instance²: $0.00000350	$0.00000350 / GiB-second beyond free tier Free: First 257,142 GiB-seconds free per month CUD¹: $0.000002905 If idle min instance²: $0.00000350	$0.40 / million requests beyond free tier³ Free: 2 million requests free per month CUD¹: $0.332

If you pay in a currency other than USD, the prices listed in your currency on Cloud Platform SKUs apply.

⁰ FlexCUD refers to Compute Flexible Committed Use Discounts.

¹ CUD refers to committed use discounts.

² idle min instance refers to idle billable time for instances kept warm using minimum instances. Idle instances that are not minimum instances are not charged.

³ Requests are only billed when they reach the container after successfully being authenticated, requests denied by IAM policy are not billed.

GPU pricing for regions in Tier 1

GPU type	Price
NVIDIA-L4	$0.000233 / GPU-second

GPU pricing for regions in Tier 2

GPU type	Price
NVIDIA-L4	$0.0002796 / GPU-second

The above GPU prices are in addition to the CPU and RAM prices for your Cloud Run instance, and uses CPU always allocated billable instance time

Regional price tiers

Subject to Tier 1 pricing

asia-east1 (Taiwan)
asia-northeast1 (Tokyo)
asia-northeast2 (Osaka)
asia-south1 (Mumbai, India)
europe-north1 (Finland) Low CO₂
europe-southwest1 (Madrid) Low CO₂
europe-west1 (Belgium) Low CO₂
europe-west4 (Netherlands) Low CO₂
europe-west8 (Milan)
europe-west9 (Paris) Low CO₂
me-west1 (Tel Aviv)
us-central1 (Iowa) Low CO₂
us-east1 (South Carolina)
us-east4 (Northern Virginia)
us-east5 (Columbus)
us-south1 (Dallas) Low CO₂
us-west1 (Oregon) Low CO₂

Subject to Tier 2 pricing

africa-south1 (Johannesburg)
asia-east2 (Hong Kong)
asia-northeast3 (Seoul, South Korea)
asia-southeast1 (Singapore)
asia-southeast2 (Jakarta)
asia-south2 (Delhi, India)
australia-southeast1 (Sydney)
australia-southeast2 (Melbourne)
europe-central2 (Warsaw, Poland)
europe-west10 (Berlin) Low CO₂
europe-west12 (Turin)
europe-west2 (London, UK) Low CO₂
europe-west3 (Frankfurt, Germany) Low CO₂
europe-west6 (Zurich, Switzerland) Low CO₂
me-central1 (Doha)
me-central2 (Dammam)
northamerica-northeast1 (Montreal) Low CO₂
northamerica-northeast2 (Toronto) Low CO₂
southamerica-east1 (Sao Paulo, Brazil) Low CO₂
southamerica-west1 (Santiago, Chile) Low CO₂
us-west2 (Los Angeles)
us-west3 (Salt Lake City)
us-west4 (Las Vegas)

Billable instance time

The billable time aggregated from all Cloud Run instances is exposed as a Cloud Monitoring metric. See container/billable_instance_time metric for more details.

Billable instance time is rounded up to the nearest 100 milliseconds and depends on the CPU allocation configuration of your Cloud Run service:

Billable instance time

Billable instance time for services with CPU allocated only during request processing

By default, Cloud Run only charges for the CPU and memory allocated to an instance when:

The instance is starting.
The instance is gracefully shutting down (handling the SIGTERM signal).
At least one request is being processed by the instance. Billable instance time begins with the start of the first request and ends at the end of the last request, as shown in the following diagram:

If you set a minimum number of instances, you are also billed at a different "idle" rate when these instances are not processing requests. See the table above.

Billable instance time for services with CPU always allocated

When you opt-into having CPU always allocated, you are billed for the entire lifetime any Cloud Run container instances: from the time the container is started to when it is terminated, with a minimum of 1 minute.

Billable instance time for Cloud Run jobs

Cloud Run jobs are billed at the CPU always allocated rate, for the entire lifetime of any instance started, with a minimum of 1 minute.

Pricing Examples

Example 1: Web service serving 10 million requests per month

Let's assume that you deployed a Cloud Run Service with CPU only allocated during request in us-central1. Your service receives 10 million requests per month with an average per-request latency of 1 second. This service is configured with 1 vCPU, 512 MiB and 80 maximum concurrent requests per instance.

The monthly pricing estimate for this workload is $3.20 ($6.36 excluding the vCPU/Memory free tier). You can either read below for the fine-grained calculations or view the summarized results via the Pricing calculator.

Request charges

Cloud Run requests are billed at $0.40 / million requests beyond the free tier. The first 2 million requests per month are free. The request charges for this workload are as follows:

(10 million requests) * ($0.40 / million requests) = $4.00

When including the free tier, the request charges are as follows:

$4.00 - ((2 million requests) * ($0.40 / million requests)) = $3.20

vCPU charges

Cloud Run vCPU is billed at $0.00002400 / vCPU-second beyond the free tier in us-central1. The first 180,000 vCPU-seconds per month are free. The total vCPU-seconds consumed per month is calculated as follows:

(1 vCPU per instance) * (10 million requests) * (1 second per-request latency) / (80 maximum concurrent requests) = 125,000 vCPU-seconds

The vCPU charges for this workload are as follows:

(125,000 vCPU-seconds) * ($0.00002400 / vCPU-second) = $3.00

When including the free tier, the vCPU charges are as follows:

$3.00 - ((180,000 vCPU-seconds) * ($0.00002400 / vCPU-second)) = $0.00

Memory charges

Cloud Run Memory is billed at $0.00000250 / GiB-second beyond the free tier in us-central1. The first 360,000 GiB-seconds per month are free. The total GiB-seconds consumed per month is calculated as follows:

(0.5 GiB per instance) * (10 million requests) * (1 second per-request latency) / (80 maximum concurrent requests) = 62,500 GiB-seconds

The Memory charges for this workload are as follows:

(62,500 GiB-seconds) * ($0.00000250 / GiB-second) = $0.15625

When including the free tier, the Memory charges are as follows:

$0.15625 - ((360,000 GiB-seconds) * ($0.00000250 / GiB-second)) = $0.00

Total charge

When excluding the vCPU & Memory free tiers, the total charge for this workload is as follows:

($3.20 Request charges) + ($3.00 vCPU charges) + ($0.15625 Memory charges) = $6.35625

When including the vCPU & Memory free tiers, the total charge for this workload is as follows:

($3.20 Request charges) + ($0.00 vCPU charges) + ($0.00 Memory charges) = $3.20

Example 2: Public API serving 50 million requests per month with reduced cold start

Let's assume that you deployed a Cloud Run Service with CPU only allocated during request in us-central1 with 2 min instances. Your service receives 50 million requests per month with an average per-request latency of 1 second. This service is configured with 1 vCPU, 2 GiB and 80 maximum concurrent requests per instance.

The monthly pricing estimate for this workload is $66.84 ($72.06 excluding the vCPU/Memory free tier). You can either read below for the fine-grained calculations or view the summarized results via the Pricing calculator.

At this scale, it is cheaper to configure CPU only allocated during request since the underlying instances are active (processing requests) for just ~11.9% of the month.

If this workload enabled CPU always allocated, the monthly pricing estimate jumps up to $119.84. You can view the summarized results with CPU always allocated via the Pricing calculator. Please reference this documentation for additional Cloud Run cost optimization guidance.

Request charges

Cloud Run requests are billed at $0.40 / million requests beyond the free tier. The first 2 million requests per month are free. The request charges for this workload are as follows:

(50 million requests) * ($0.40 / million requests) = $20.00

When including the free tier, the request charges are as follows:

$20.00 - ((2 million requests) * ($0.40 / million requests)) = $19.20

vCPU charges

(1 vCPU per instance) * (50 million requests) * (1 second per-request latency) / (80 maximum concurrent requests) = 625,000 vCPU-seconds

The vCPU charges for this workload are as follows:

(625,000 vCPU-seconds) * ($0.00002400 / vCPU-second) = $15.00

When including the free tier, the vCPU charges are as follows:

$15.00 - ((180,000 vCPU-seconds) * ($0.00002400 / vCPU-second)) = $10.68

Memory charges

(2 GiB per instance) * (50 million requests) * (1 second per-request latency) / (80 maximum concurrent requests) = 1,250,000 GiB-seconds

The Memory charges for this workload are as follows:

(1,250,000 GiB-seconds) * ($0.00000250 / GiB-second) = $3.125

When including the free tier, the Memory charges are as follows:

$3.125 - ((360,000 GiB-seconds) * ($0.00000250 / GiB-second)) = $2.225

Min Instance vCPU charges

Cloud Run Idle Min Instance vCPU is billed at $0.00000250 / vCPU-second beyond the free tier in us-central1. The two min instances were alive for the entire month but are only billed when idle. For simplicity, let's assume that there are 365 days per year and 86,400 seconds per day. The two min instances consumed the following amount of idle vCPU-seconds per month:

(365 days) / (12 months) * (86,400 seconds) * (2 instances) * (1 vCPU per instance) - (625,000 vCPU-seconds active) = 4,631,000 vCPU-seconds

The idle min instance vCPU charges for this workload are as follows:

(4,631,000 vCPU-seconds) * ($0.00000250 / vCPU-second) = $11.5775

Min Instance Memory charges

Cloud Run Idle Min Instance Memory is billed at $0.00000250 / GiB-second beyond the free tier in us-central1. The two min instances were alive for the entire month but are only billed when idle. For simplicity, let's assume that there are 365 days per year and 86,400 seconds per day. The two min instances consumed the following amount of idle GiB-seconds per month:

(365 days) / (12 months) * (86,400 seconds) * (2 instances) * (2 GiB per instance) - (1,250,000 vCPU-seconds active) = 9,262,000 GiB-seconds

The idle min instance Memory charges for this workload are as follows:

(9,262,000 GiB-seconds) * ($0.00000250 / GiB-second) = $23.155

Total charge

When excluding the vCPU & Memory free tiers, the total charge for this workload is as follows:

($19.20 Request charges) + ($15.00 vCPU charges) + ($3.125 Memory charges) + ($11.5775 Min Instance vCPU charge) + ($23.155 Min Instance Memory charge) = $72.0575

When including the vCPU & Memory free tiers, the total charge for this workload is as follows:

($19.20 Request charges) + ($10.68 vCPU charges) + ($2.225 Memory charges) + ($11.5775 Min Instance vCPU charge) + ($23.155 Min Instance Memory charge) = $66.8375

Example 3: Mobile backend serving 1 billion requests per month

Let's assume that you deployed a Cloud Run Service with CPU always allocated in europe-west3. Your service receives 1 billion requests per month with an average per-request latency of 525.6 milliseconds. This service is configured with 1 vCPU, 512 MiB and 100 maximum concurrent requests per instance.

The monthly pricing estimate for this workload is $119.84 ($64.71 with a 3-year FlexCUD). You can either read below for the fine-grained calculations or view the summarized results via the Pricing calculator.

At this scale, it is cheaper to configure CPU always allocated since:

The underlying instances are always active (processing requests).
When CPU always allocated is enabled, requests are free.
When CPU always allocated is enabled, vCPU/Memory allocation time is cheaper.

If this workload enabled CPU allocated only during request processing, the monthly pricing estimate jumps up to $585.00. You can view the summarized results with CPU allocated only during request processing via the Pricing calculator. Please reference this documentation for additional Cloud Run cost optimization guidance.

Request charges

When CPU always allocated is enabled, Cloud Run requests are free.

Even though requests are free for this Cloud Run Service, the auto-scaling behavior is still driven by incoming requests. If we assume that there are 365 days per year and 86,400 seconds per day, we are able to calculate the number of concurrent requests as follows:

(1 billion requests) * (0.5256 second per-request latency) / ((365 days) / (12 months) * (86,400 seconds)) = 200 concurrent requests

As each instance is able to handle 100 concurrent requests, this Cloud Run Service would need 2 instances to handle 200 concurrent requests.

vCPU charges

Cloud Run vCPU is billed at $0.00002160 / vCPU-second beyond the free tier in europe-west3. The total vCPU-seconds consumed per month is calculated as follows:

(2 instances) * (1 vCPU per instance) * (365 days) / (12 months) * (86,400 seconds) = 5,256,000 vCPU-seconds

The vCPU charges for this workload are as follows:

(5,256,000 vCPU-seconds) * ($0.00002160 / vCPU-second) = $113.5296

With a 3-year FlexCUD, the vCPU charges are as follows:

(5,256,000 vCPU-seconds) * ($0.000011664 / vCPU-second) = $61.305984

Memory charges

Cloud Run Memory is billed at $0.00000240 / GiB-second beyond the free tier in europe-west3. The total GiB-seconds consumed per month is calculated as follows:

(2 instances) * (0.5 GiB per instance) * (365 days) / (12 months) * (86,400 seconds) = 2,628,000 GiB-seconds

The Memory charges for this workload are as follows:

(2,628,000 GiB-seconds) * ($0.00000240 / GiB-second) = $6.3072

With a 3-year FlexCUD, the Memory charges are as follows:

(2,628,000 GiB-seconds) * ($0.000001296 / GiB-second) = $3.405888

Total charge

The total charge for this workload is as follows:

($0.00 Request charges) + ($113.5296 vCPU charges) + ($6.3072 Memory charges) = $119.8368

The total charge for this workload (with a 3-year FlexCUD) is as follows:

($0.00 Request charges) + ($61.305984 vCPU charges) + ($3.405888 Memory charges) = $64.711872

Example 4: Process records from a database at the beginning of each hour

Let's assume that you deployed a Cloud Run Job in asia-east1. Your Job is configured with 2 vCPU, 4 GiB, is triggered once per hour, contains 10 tasks and each task requires 2 minutes to complete.

The monthly pricing estimate for this workload is $33.32 ($38.54 excluding the vCPU/Memory free tier). You can either read below for the fine-grained calculations or view the summarized results via the Pricing calculator.

Request charges

Cloud Run Jobs does not support ingress so this resource does not generate any request charges.

vCPU charges

Cloud Run vCPU is billed at $0.00001800 / vCPU-second beyond the free tier in us-central1. The first 240,000 vCPU-seconds per month are free. The total vCPU-seconds consumed per month is calculated as follows:

(730 hours per month) * (1 execution per hour) * (10 tasks per execution) * (120 seconds per task execution) * (2 vCPU per instance) = 1,752,000 vCPU-seconds

The vCPU charges for this workload are as follows:

(1,752,000 vCPU-seconds) * ($0.00001800 / vCPU-second) = $31.536

When including the free tier, the vCPU charges are as follows:

$31.536 - ((240,000 vCPU-seconds) * ($0.00001800 / vCPU-second)) = $27.216

Memory charges

Cloud Run Memory is billed at $0.00000200 / GiB-second beyond the free tier in us-central1. The first 450,000 GiB-seconds per month are free. The total GiB-seconds consumed per month is calculated as follows:

(730 hours per month) * (1 execution per hour) * (10 tasks per execution) * (120 seconds per task execution) * (4 GiB per instance) = 3,504,000 GiB-seconds

The Memory charges for this workload are as follows:

(3,504,000 GiB-seconds) * ($0.00000200 / GiB-second) = $7.008

When including the free tier, the Memory charges are as follows:

$7.008 - ((450,000 GiB-seconds) * ($0.00000200 / GiB-second)) = $6.108

Total charge

When excluding the vCPU & Memory free tiers, the total charge for this workload is as follows:

($0.00 Request charges) + ($31.536 vCPU charges) + ($7.008 Memory charges) = $38.544

When including the vCPU & Memory free tiers, the total charge for this workload is as follows:

($0.00 Request charges) + ($27.216 vCPU charges) + ($6.108 Memory charges) = $33.324

Example 5: Backend service performing AI inference using LLMs for 10 million requests per month

Let's assume that you deployed a Cloud Run Service with CPU always allocated and Nvidia L4 GPU in us-central1. Your service receives 10 million requests per month with an average per-request latency of 2 seconds. This service is configured with 4 vCPU, 16 GiB and 10 maximum concurrent requests per instance.

The monthly pricing estimate for this workload is $885.64 ($809.11 with a 1-year FlexCUD). You can either read below for the fine-grained calculations or view the summarized results via the Pricing calculator.

Request charges

When CPU always allocated is enabled, Cloud Run requests are free.

(10 million requests) * (2 second per-request latency) / ((365 days) / (12 months) * (86,400 seconds)) = 7.61 concurrent requests

As each instance is able to handle 10 concurrent requests, this Cloud Run Service would need 1 instances to handle 7.61 concurrent requests.

vCPU charges

(365 days) / (12 months) * (86,400 seconds) * (1 instance) * (4 vCPU per instance) = 10,512,000 vCPU-seconds

The vCPU charges for this workload are as follows:

(10,512,000 vCPU-seconds) * ($0.00001800 / vCPU-second) = $189.216

With a 1-year FlexCUD, the vCPU charges are as follows:

(10,512,000 vCPU-seconds) * ($0.00001296 / vCPU-second) = $136.23552

Memory charges

(365 days) / (12 months) * (86,400 seconds) * (1 instance) * (16 GiB per instance) = 42,048,000 GiB-seconds

The Memory charges for this workload are as follows:

(42,048,000 GiB-seconds) * ($0.00000200 / GiB-second) = $84.096

With a 1-year FlexCUD, the Memory charges are as follows:

(42,048,000 GiB-seconds) * ($0.00000144 / GiB-second) = $60.54912

GPU charges

Cloud Run Nvidia L4 GPU is billed at $0.000233 / GPU-second in us-central1. The total GPU-seconds consumed per month is calculated as follows:

(365 days) / (12 months) * (86,400 seconds) * (1 instance) = 2,628,000 GPU-seconds

The GPU charges for this workload are as follows:

(2,628,000 GPU-seconds) * ($0.000233 / GPU-second) = $612.324

Total charge

The total charge for this workload is as follows:

($0.00 Request charges) + ($189.216 vCPU charges) + ($84.096 Memory charges) + ($612.324 GPU charges) = $885.636

The total charge for this workload (with a 1-year FlexCUD) is as follows:

($0.00 Request charges) + ($136.23552 vCPU charges) + ($60.54912 Memory charges) + ($612.324 GPU charges) = $809.10864

Source deployments and functions

Cloud Run deployments from source and functions operate as Cloud Run services and are billed accordingly based on the conditions above.

When deploying from source or creating a function, Cloud Run uses Cloud Build to convert your source code or function into an executable image. The images are created in Cloud Build and stored in Artifact Registry. Artifact Registry has its own free tier of storage. If you deploy your source code or function to Artifact Registry and exceed the Artifact Registry free tier usage, you will incur charges for deploying your functions, even when your use of Cloud Run falls within the free tier.

Pricing for Cloud Build and Artifact Registry is not included in Cloud Run pricing. To review pricing for these services, see the Cloud Build Pricing and Artifact Registry Pricing pages.

Cloud Run uses Eventarc for event delivery. Pricing for Eventarc is not included in the Cloud Run pricing above. Review pricing on the Eventarc Pricing page.

Pricing for Cloud Run functions (1st gen) functions is not included in Cloud Run pricing. Review the pricing in Cloud Run functions (1st gen) pricing.

What's next

Read the Cloud Run documentation.
Get started with Cloud Run.
Try the Pricing calculator.
Learn about Cloud Run solutions and use cases.

Request a custom quote

With Google Cloud's pay-as-you-go pricing, you only pay for the services you use. Connect with our sales team to get a custom quote for your organization.

Contact sales