Quotas and limits

This page lists the default quotas that apply to Vertex AI. Your quotas might differ from the defaults if you have previously requested higher quotas.

For more information about quotas in Google Cloud, including how to request an increase, see the quotas documentation.

Request quotas

The following quotas apply to Vertex AI requests for a given project and region. For example, in a single project, you can have up to 30,000 online prediction requests per minute in one region and another 30,000 online prediction requests per minute in another region.

Request quota	Value
Resource management requests^* per minute	600
Job or long-running operation requests per minute	60
Online prediction requests per minute^+	30,000
Online prediction request throughput per minute	1.5 GB
Online explanation requests per minute	600
Vertex AI Vizier requests per minute	6,000
Vertex AI Feature Store online serving requests per minute	300,000
Vertex ML Metadata requests per minute	12,000

* Resource management requests include any request that is not a job, long-running operation, online prediction request, or Vertex AI Vizier request.

+ This quota applies for public endpoints only. Private endpoints have unlimited requests per minute.

Jobs or long-running operations include the following requests:

• Creating or deleting a dataset
• Importing or exporting data to or from a dataset
• Creating an endpoint
• Creating or deleting a custom job
• Creating or deleting a data labeling job
• Creating or deleting a hyper parameter tuning job
• Creating or deleting a batch prediction job
• Creating or deleting a model
• Uploading, deleting, or exporting a model

AutoML model quotas

The following quotas apply to each data type and objective for a given project and region. For example, in a particular project and region, you can deploy 10 AutoML image classification models and 10 AutoML image object detection models for a total of 20 deployed models.

Image

Classification

Quota	Value
Concurrent training jobs	5
Concurrent training jobs with explainable AI	2
Concurrent batch prediction jobs	5
Concurrent model deployment jobs	5
Concurrent model undeployment jobs	5
Number of deployed models	10

Object detection

Quota	Value
Concurrent training jobs	5
Concurrent batch prediction jobs	5
Number of deployed models	10

Tabular

Quota	Value
Concurrent training jobs	5
Concurrent batch prediction jobs	5
Number of deployed models	30

Text

Classification

Quota	Value
Concurrent training jobs	5
Concurrent batch prediction jobs	5
Number of deployed models	10

Entity extraction

Quota	Value
Concurrent training jobs	5
Concurrent batch prediction jobs	5
Number of deployed models	10

Sentiment analysis

Quota	Value
Concurrent training jobs	5
Concurrent batch prediction jobs	5
Number of deployed models	10

Video

Action Recognition

Quota	Value
Concurrent training jobs	5
Concurrent batch prediction jobs	5

Classification

Quota	Value
Concurrent training jobs	5
Concurrent batch prediction jobs	5

Object tracking

Quota	Value
Concurrent training jobs	5
Concurrent batch prediction jobs	5

AutoML Video Intelligence specific limits

The following describes current restrictions on the use of AutoML Video Intelligence.

Type of limits	Value
Maximum video length	3 hours
Maximum video file size	50GB
Minimum labels per dataset	2
Minimum videos per label	10 (1000 is recommended)
Batch input CSV file size	Maximum: 100MB
Number of video segments in batch input	Maximum: 1,000

Custom-trained model quotas

The following quotas apply to Vertex AI custom-trained models for a given project and region.

Note: The number of CPUs or GPUs are not counted against Compute Engine quotas. They are counted separately and cannot be used together.

Training

Number of concurrent N1 + E2 CPUs for training, per region
Region	Value
us-west1	2,200
us-west2	20
us-west3	2,200
us-west4	20
us-central1	2,200
us-east1	2,200
us-east4	20
us-south1	450
northamerica-northeast1	2,200
northamerica-northeast2	20
southamerica-east1	20
europe-west2	2,200
europe-west1	2,200
europe-west4	2,200
europe-west6	20
europe-west3	2,200
europe-central2	450
europe-west9	450
asia-south1	2,200
asia-southeast1	2,200
asia-southeast2	2,200
asia-east2	2,200
asia-east1	2,200
asia-northeast1	2,200
australia-southeast1	2,200
asia-northeast3	2,200
me-west1	450

Number of concurrent A2 CPUs for training, per region
Region	Value
us-west1	Not available
us-west2	Not available
us-west3	Not available
us-west4	Not available
us-central1	96
us-east1	Not available
us-east4	Not available
us-south1	Not available
northamerica-northeast1	Not available
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	Not available
europe-west1	Not available
europe-west4	96
europe-west6	Not available
europe-west3	Not available
europe-central2	Not available
europe-west9	Not available
asia-south1	Not available
asia-southeast1	96
asia-southeast2	Not available
asia-east2	Not available
asia-east1	Not available
asia-northeast1	Not available
australia-southeast1	Not available
asia-northeast3	Not available
me-west1	Not available

Number of concurrent N2 CPUs for training, per region
Region	Value
us-west1	20
us-west2	20
us-west3	20
us-west4	20
us-central1	450
us-east1	20
us-east4	20
us-south1	20
northamerica-northeast1	20
northamerica-northeast2	20
southamerica-east1	20
europe-west2	20
europe-west1	20
europe-west4	450
europe-west6	20
europe-west3	20
europe-central2	20
europe-west9	450
asia-south1	20
asia-southeast1	20
asia-southeast2	20
asia-east2	20
asia-east1	450
asia-northeast1	20
australia-southeast1	20
asia-northeast3	20
me-west1	20

Number of concurrent C2 CPUs for training, per region
Region	Value
us-west1	20
us-west2	20
us-west3	20
us-west4	20
us-central1	450
us-east1	20
us-east4	20
us-south1	20
northamerica-northeast1	20
northamerica-northeast2	20
southamerica-east1	20
europe-west2	20
europe-west1	20
europe-west4	450
europe-west6	20
europe-west3	20
europe-central2	20
europe-west9	450
asia-south1	20
asia-southeast1	20
asia-southeast2	20
asia-east2	20
asia-east1	450
asia-northeast1	20
australia-southeast1	20
asia-northeast3	20
me-west1	20

Number of concurrent A100 GPUs for training, per region
Region	Value
us-west1	Not available
us-west2	Not available
us-west3	Not available
us-west4	Not available
us-central1	8
us-east1	Not available
us-east4	Not available
us-south1	Not available
northamerica-northeast1	Not available
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	Not available
europe-west1	Not available
europe-west4	8
europe-west6	Not available
europe-west3	Not available
europe-central2	Not available
europe-west9	Not available
asia-south1	Not available
asia-southeast1	8
asia-southeast2	Not available
asia-east2	Not available
asia-east1	Not available
asia-northeast1	Not available
australia-southeast1	Not available
asia-northeast3	Not available
me-west1	Not available

Number of concurrent A100 80GB GPUs for training, per region
Region	Value
us-west1	Not available
us-west2	Not available
us-west3	Not available
us-west4	Not available
us-central1	0
us-east1	Not available
us-east4	0
us-south1	Not available
northamerica-northeast1	Not available
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	Not available
europe-west1	Not available
europe-west4	Not available
europe-west6	Not available
europe-west3	Not available
europe-central2	Not available
europe-west9	Not available
asia-south1	Not available
asia-southeast1	Not available
asia-southeast2	Not available
asia-east2	Not available
asia-east1	Not available
asia-northeast1	Not available
australia-southeast1	Not available
asia-northeast3	Not available
me-west1	Not available

If interested, please see the quotas documentation.

Number of concurrent K80 GPUs for training, per region
Region	Value
us-west1	30
us-west2	Not available
us-west3	Not available
us-west4	Not available
us-central1	56
us-east1	30
us-east4	Not available
us-south1	Not available
northamerica-northeast1	Not available
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	Not available
europe-west1	30
europe-west4	Not available
europe-west6	Not available
europe-west3	Not available
europe-central2	Not available
europe-west9	Not available
asia-south1	Not available
asia-southeast1	Not available
asia-southeast2	Not available
asia-east2	Not available
asia-east1	56
asia-northeast1	Not available
australia-southeast1	Not available
asia-northeast3	Not available
me-west1	Not available

Number of concurrent P100 GPUs for training, per region
Region	Value
us-west1	30
us-west2	Not available
us-west3	Not available
us-west4	Not available
us-central1	56
us-east1	30
us-east4	Not available
us-south1	Not available
northamerica-northeast1	Not available
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	Not available
europe-west1	30
europe-west4	Not available
europe-west6	Not available
europe-west3	Not available
europe-central2	Not available
europe-west9	Not available
asia-south1	Not available
asia-southeast1	Not available
asia-southeast2	Not available
asia-east2	Not available
asia-east1	30
asia-northeast1	Not available
australia-southeast1	6
asia-northeast3	Not available
me-west1	Not available

Number of concurrent P4 GPUs for training, per region
Region	Value
us-west1	Not available
us-west2	6
us-west3	Not available
us-west4	Not available
us-central1	6
us-east1	Not available
us-east4	1
us-south1	Not available
northamerica-northeast1	6
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	Not available
europe-west1	Not available
europe-west4	6
europe-west6	Not available
europe-west3	Not available
europe-central2	Not available
europe-west9	Not available
asia-south1	Not available
asia-southeast1	6
asia-southeast2	Not available
asia-east2	Not available
asia-east1	Not available
asia-northeast1	Not available
australia-southeast1	6
asia-northeast3	Not available
me-west1	Not available

Number of concurrent T4 GPUs for training, per region
Region	Value
us-west1	2
us-west2	Not available
us-west3	Not available
us-west4	Not available
us-central1	12
us-east1	2
us-east4	Not available
us-south1	Not available
northamerica-northeast1	Not available
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	6
europe-west1	Not available
europe-west4	2
europe-west6	Not available
europe-west3	0
europe-central2	Not available
europe-west9	Not available
asia-south1	6
asia-southeast1	1
asia-southeast2	Not available
asia-east2	Not available
asia-east1	Not available
asia-northeast1	6
australia-southeast1	Not available
asia-northeast3	1
me-west1	Not available

Number of concurrent V100 GPUs for training, per region
Region	Value
us-west1	6
us-west2	Not available
us-west3	Not available
us-west4	Not available
us-central1	6
us-east1	Not available
us-east4	Not available
us-south1	Not available
northamerica-northeast1	Not available
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	Not available
europe-west1	Not available
europe-west4	6
europe-west6	Not available
europe-west3	Not available
europe-central2	Not available
europe-west9	Not available
asia-south1	Not available
asia-southeast1	Not available
asia-southeast2	Not available
asia-east2	Not available
asia-east1	6
asia-northeast1	Not available
australia-southeast1	Not available
asia-northeast3	Not available
me-west1	Not available

Number of concurrent TPU V2 cores for training, per region
Region	Value
us-west1	Not available
us-west2	Not available
us-west3	Not available
us-west4	Not available
us-central1	8
us-east1	Not available
us-east4	Not available
us-south1	Not available
northamerica-northeast1	Not available
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	Not available
europe-west1	Not available
europe-west4	8
europe-west6	Not available
europe-west3	Not available
europe-central2	Not available
europe-west9	Not available
asia-south1	Not available
asia-southeast1	Not available
asia-southeast2	Not available
asia-east2	Not available
asia-east1	8
asia-northeast1	Not available
australia-southeast1	Not available
asia-northeast3	Not available
me-west1	Not available

Number of concurrent TPU V2 pod cores for training, per region
Region	Value
us-west1	Not available
us-west2	Not available
us-west3	Not available
us-west4	Not available
us-central1	Not available
us-east1	Not available
us-east4	Not available
us-south1	Not available
northamerica-northeast1	Not available
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	Not available
europe-west1	Not available
europe-west4	Not available
europe-west6	Not available
europe-west3	Not available
europe-central2	Not available
europe-west9	Not available
asia-south1	Not available
asia-southeast1	Not available
asia-southeast2	Not available
asia-east2	Not available
asia-east1	Not available
asia-northeast1	Not available
australia-southeast1	Not available
asia-northeast3	Not available
me-west1	Not available

Number of concurrent TPU V3 cores for training, per region
Region	Value
us-west1	Not available
us-west2	Not available
us-west3	Not available
us-west4	Not available
us-central1	8
us-east1	Not available
us-east4	Not available
us-south1	Not available
northamerica-northeast1	Not available
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	Not available
europe-west1	Not available
europe-west4	8
europe-west6	Not available
europe-west3	Not available
europe-central2	Not available
europe-west9	Not available
asia-south1	Not available
asia-southeast1	Not available
asia-southeast2	Not available
asia-east2	Not available
asia-east1	8
asia-northeast1	Not available
australia-southeast1	Not available
asia-northeast3	Not available
me-west1	Not available

Number of concurrent TPU V3 pod cores for training, per region
Region	Value
us-west1	Not available
us-west2	Not available
us-west3	Not available
us-west4	Not available
us-central1	Not available
us-east1	Not available
us-east4	Not available
us-south1	Not available
northamerica-northeast1	Not available
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	Not available
europe-west1	Not available
europe-west4	Not available
europe-west6	Not available
europe-west3	Not available
europe-central2	Not available
europe-west9	Not available
asia-south1	Not available
asia-southeast1	Not available
asia-southeast2	Not available
asia-east2	Not available
asia-east1	Not available
asia-northeast1	Not available
australia-southeast1	Not available
asia-northeast3	Not available
me-west1	Not available

HDD usage (GB) during training, per region
Region	Value
us-west1	180,000
us-west2	3,600
us-west3	180,000
us-west4	3,600
us-central1	180,000
us-east1	180,000
us-east4	3,600
us-south1	180,000
northamerica-northeast1	180,000
northamerica-northeast2	3,600
southamerica-east1	3,600
europe-west2	180,000
europe-west1	180,000
europe-west4	180,000
europe-west6	3,600
europe-west3	180,000
europe-central2	180,000
europe-west9	180,000
asia-south1	180,000
asia-southeast1	180,000
asia-southeast2	180,000
asia-east2	180,000
asia-east1	180,000
asia-northeast1	180,000
australia-southeast1	180,000
asia-northeast3	180,000
me-west1	180,000

SSD usage (GB) during training, per region
Region	Value
us-west1	75,000
us-west2	450
us-west3	75,000
us-west4	450
us-central1	75,000
us-east1	75,000
us-east4	450
us-south1	75,000
northamerica-northeast1	75,000
northamerica-northeast2	450
southamerica-east1	450
europe-west2	75,000
europe-west1	75,000
europe-west4	75,000
europe-west6	450
europe-west3	75,000
europe-central2	75,000
europe-west9	75,000
asia-south1	75,000
asia-southeast1	75,000
asia-southeast2	75,000
asia-east2	75,000
asia-east1	75,000
asia-northeast1	75,000
australia-southeast1	75,000
asia-northeast3	75,000
me-west1	75,000

Serving

Quota	Value
Number of deployed custom models	100

Number of concurrent CPUs for serving, per region
Region	Value
us-west1	2,200
us-west2	2,200
us-west3	2,200
us-west4	16
us-central1	2,200
us-east1	2,200
us-east4	2,200
us-south1	450
northamerica-northeast1	2,200
northamerica-northeast2	450
southamerica-east1	16
europe-west2	2,200
europe-west1	2,200
europe-west4	2,200
europe-west6	2,200
europe-west3	2,200
europe-central2	450
europe-west9	16
asia-south1	2,200
asia-southeast1	2,200
asia-southeast2	2,200
asia-east2	2,200
asia-east1	2,200
asia-northeast1	2,200
australia-southeast1	2,200
asia-northeast3	2,200
me-west1	450

Number of concurrent K80 GPUs for serving, per region
Region	Value
us-west1	Not available
us-west2	Not available
us-west3	Not available
us-west4	Not available
us-central1	56
us-east1	30
us-east4	Not available
us-south1	Not available
northamerica-northeast1	Not available
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	Not available
europe-west1	30
europe-west4	Not available
europe-west6	Not available
europe-west3	Not available
europe-central2	Not available
europe-west9	Not available
asia-south1	Not available
asia-southeast1	Not available
asia-southeast2	Not available
asia-east2	Not available
asia-east1	56
asia-northeast1	Not available
australia-southeast1	Not available
asia-northeast3	Not available
me-west1	Not available

Number of concurrent P100 GPUs for serving, per region
Region	Value
us-west1	30
us-west2	Not available
us-west3	Not available
us-west4	Not available
us-central1	56
us-east1	30
us-east4	Not available
us-south1	Not available
northamerica-northeast1	Not available
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	Not available
europe-west1	30
europe-west4	Not available
europe-west6	Not available
europe-west3	Not available
europe-central2	Not available
europe-west9	Not available
asia-south1	Not available
asia-southeast1	Not available
asia-southeast2	Not available
asia-east2	Not available
asia-east1	30
asia-northeast1	Not available
australia-southeast1	Not available
asia-northeast3	Not available
me-west1	Not available

Number of concurrent P4 GPUs for serving, per region
Region	Value
us-west1	Not available
us-west2	6
us-west3	Not available
us-west4	Not available
us-central1	6
us-east1	Not available
us-east4	6
us-south1	Not available
northamerica-northeast1	6
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	Not available
europe-west1	Not available
europe-west4	6
europe-west6	Not available
europe-west3	Not available
europe-central2	Not available
europe-west9	Not available
asia-south1	Not available
asia-southeast1	6
asia-southeast2	Not available
asia-east2	Not available
asia-east1	Not available
asia-northeast1	Not available
australia-southeast1	6
asia-northeast3	Not available
me-west1	Not available

Number of concurrent T4 GPUs for serving, per region
Region	Value
us-west1	12
us-west2	Not available
us-west3	Not available
us-west4	Not available
us-central1	12
us-east1	12
us-east4	Not available
us-south1	Not available
northamerica-northeast1	Not available
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	12
europe-west1	Not available
europe-west4	12
europe-west6	Not available
europe-west3	0
europe-central2	Not available
europe-west9	Not available
asia-south1	6
asia-southeast1	6
asia-southeast2	Not available
asia-east2	Not available
asia-east1	6
asia-northeast1	6
australia-southeast1	Not available
asia-northeast3	6
me-west1	Not available

Number of concurrent V100 GPUs for serving, per region
Region	Value
us-west1	6
us-west2	Not available
us-west3	Not available
us-west4	Not available
us-central1	6
us-east1	Not available
us-east4	Not available
us-south1	Not available
northamerica-northeast1	Not available
northamerica-northeast2	Not available
southamerica-east1	Not available
europe-west2	Not available
europe-west1	Not available
europe-west4	6
europe-west6	Not available
europe-west3	Not available
europe-central2	Not available
europe-west9	Not available
asia-south1	Not available
asia-southeast1	Not available
asia-southeast2	Not available
asia-east2	Not available
asia-east1	Not available
asia-northeast1	Not available
australia-southeast1	Not available
asia-northeast3	Not available
me-west1	Not available

Vertex AI Feature Store

The following quotas apply to a given project and region. For example, in a single project, you can have 75 concurrent batch jobs in us-central1 and another 75 jobs in europe-west4.

Quota	Value
Online serving requests per minute	300,000
Streaming ingestion requests per minute	60,000
Streaming ingestion write throughput per minute	1.2 GB
Feature creation requests per minute	100
Online serving nodes across all featurestores	30
Concurrent batch jobs (ingestion, serving, and delete feature values combined)	75
Concurrent requests to delete feature values	1
Entity types across all featurestores	75

Vertex AI Feature Store also has the following limits. Note that, unlike quotas, you cannot request an increase to a limit.

Limit	Value
Storage limit for an online serving node	5 TB
Total data in the offline store	Unlimited
Features per entity type	5,000
Number of create, update, and delete featurestore requests per day per project per region	500
For streaming ingestion, the size per request	1 MB
For streaming read, the number of entities that can be included per request	100
For batch import, the number of files that can be included per request	5,000 for Avro or 500 for CSV
For batch serving and exports, the number of features you can request	5,000
Data retention (oldest feature value timestamp after which the values are deleted)	4,000 days from the current date
For batch ingestion and streaming ingestion, the oldest timestamp for which feature data can be ingested	4,000 days from current date

Note: Vertex AI Feature Store uses column-partitioned tables from BigQuery. Your project can make up to 30,000 partition modifications per day for a column-partitioned table in BigQuery. If you exceed this limit, Vertex AI returns an error. For more information about this error, see Number of partition modifications for column-partitioned tables quota errors.
Learn more about limits for BigQuery partitioned tables.

Vertex AI Matching Engine

The following quotas apply to Vertex AI Matching Engine for a given project in each region.

Quota	Value
Concurrent index creation operations	5
Concurrent index update operations	5
Number of deployed index nodes	50
Number of deployed index N2D nodes	5
Number of Index	100
Streaming Update requests per minute	6000
Streaming Update throughput(in KB) per minute	120000

Vertex AI Pipelines

The following quotas and limits apply to Vertex AI Pipelines for a given project in each region.

Quota	Value
Running pipeline tasks in parallel	600
Concurrent pipeline job*	300

* Pipeline runs beyond this limit will be queued until resources are available.

Vertex AI Pipelines also has the following limits. Note that, unlike quotas, you cannot request an increase to a limit.

Limit	Value
Number of pipeline tasks per job	10,000
Input and output artifacts per pipeline task	100
Input and output artifacts per pipeline job	10,000

Quota increases

If you want to increase any of your quotas for Vertex AI, you can use the Google Cloud console to request a quota increase.

For more information about submitting a quota increase request, see the following sections of Working with quotas:

• Requesting a higher quota

Send feedback