Amazon Personalize endpoints and quotas

The following sections contain information about Amazon Personalize guidelines, quotas, and endpoints. For adjustable quotas, you can request a quota increase using the Service Quotas console. For more information, see Requesting a quota increase.

Amazon Personalize endpoints and regions

For a list of Amazon Personalize endpoints by region, see AWS regions and endpoints in the Amazon Web Services General Reference.

Compliance

For information about Amazon Personalize compliance programs, see AWS compliance, AWS compliance programs, and AWS services in scope by compliance program.

Service quotas

Your AWS account has the following quotas for Amazon Personalize.

Resource	Quota
Item interactions
Minimum number of unique item interactions required to create a solution version or recommender. For a custom solution, you must have this many records after any filtering by event type or event value before training.	1000
For User-Personalization-v2 and Personalized-Ranking-v2 recipes, the maximum number of item interactions that are considered by a model during training.	3 billion
For all domain use cases and custom recipes other than User-Personalization-v2 or Personalized-Ranking-v2, the maximum number of item interactions that are considered by a model during training.	500 million (adjustable)
Maximum number of distinct event types combined with total number of optional metadata columns in an Item interactions dataset.	10
Maximum number of metadata columns, excluding reserved fields, in an Item interactions dataset.	5
Maximum number of characters for categorical data and impression values.	1000
Maximum amount of bulk item interactions data per dataset import job with FULL import mode.	100 GB (increases to 1TB with any increase to Item interactions considered by a model)
Maximum amount of bulk item interactions data per dataset import job with INCREMENTAL import mode.	1 GB
Minimum number of item interactions records per dataset import job with FULL or INCREMENTAL import mode.	1000
Users
Minimum number of unique users in item interactions data, with at minimum 2 item interactions each, required to create a domain recommender or custom solution version.	25
Minimum percentage of total users that must have at minimum 2 item interactions or more before you can create a domain recommender or custom solution version.	1 percent
Maximum number of metadata fields for a Users dataset.	25
Maximum number of characters for USER_ID data values.	256
Maximum number of characters for categorical data values.	1000 characters
Maximum amount of bulk user data per dataset import job with FULL import mode.	100 GB
Maximum amount of bulk user data per dataset import job with INCREMENTAL import mode.	1 GB
Items
For User-Personalization-v2 or Personalized-Ranking-v2, the maximum number of items that are considered by a model during training. These items are from both the Items and Item interactions dataset.	5 million
For all domain use cases and custom recipes other than User-Personalization-v2 and Personalized-Ranking-v2, the maximum number of items that are considered by a model during training and generating recommendations.	750,000
Maximum number of metadata fields for an Items dataset.	100
Maximum number of characters for ITEM_ID data values.	256
Maximum number of characters for categorical and non-categorical string data values.	1000 characters
Maximum number of textual fields for an Items dataset.	1
Maximum number of characters for textual data values for Chinese and Japanese languages.	7,000 characters
Maximum number of characters for textual data values for all other languages.	20,000 characters
Maximum amount of bulk items data per dataset import job with BULK import mode.	100 GB
Maximum amount of bulk item data per dataset import job with INCREMENTAL import mode.	1 GB
Actions
Maximum number of actions that are considered by a model during training and generating recommendations.	1000
Maximum number of metadata fields for an Actions dataset.	10
Maximum number of characters for ACTION_ID data values.	256
Maximum number of characters for categorical data values.	1000 characters
Maximum amount of bulk actions data per dataset import job with BULK import mode.	100 GB
Maximum amount of bulk actions data per dataset import job with INCREMENTAL import mode.	1 GB
Action interactions
Maximum number of action interactions that are considered by a model during training.	500 million
Maximum number of metadata columns, excluding reserved fields, in a Action interactions dataset.	5
Maximum amount of bulk interactions data per dataset import job with FULL import mode.	100 GB (increases to 1TB with any increase to Action item interactions considered by a model)
Maximum amount of bulk interactions data per dataset import job with INCREMENTAL import mode.	1 GB
Individual record import APIs
Maximum rate of PutEvents requests per dataset group.	1000/second
Maximum number of events in a PutEvents call.	10
Maximum size of an event.	10 KB
Maximum rate of PutActionInteractions requests per dataset group.	1000/second
Maximum number of action interaction events in a PutActionInteractions call.	10
Maximum size of an action interaction event.	10 KB
Maximum rate of PutItems requests per dataset group.	10/second
Maximum number of items in a PutItems call.	10
Maximum rate of PutUsers requests per dataset group.	10/second
Maximum number of users in a PutUsers call.	10
Maximum rate of PutActions requests per dataset group.	10/second
Maximum number of users in a PutActions call.	10
Legacy recipes
Maximum amount of combined data for Users and Items datasets for HRNN-metadata and HRNN-Coldstart recipes.	5 GB
Maximum number of cold start items the HRNN-Coldstart recipe supports to train a model (create a solution version).	80000
Minimum number of cold start items the HRNN-Coldstart recipe requires to train a model (create a solution version).	100
Filters
Total number of filters per dataset group.	30 (adjustable)
Maximum number of distinct dataset fields for a filter.	10
Total number of distinct dataset fields across all filters in a dataset group.	20
Maximum number of item interactions per user per event type considered by a filter.	100 interactions (adjustable)
Maximum number of action interactions per user per event type considered by a filter.	300 action interactions (adjustable)
GetRecommendations / GetPersonalizedRanking / GetActionRecommendations requests
Maximum transaction rate for GetRecommendations, GetActionRecommendations and GetPersonalizedRanking requests.	2500/sec
Maximum number of GetRecommendations requests per second per campaign.	500/sec
Maximum number of GetActionRecommendations requests per second per campaign.	500/sec
Maximum number of GetPersonalizedRanking requests per second per campaign.	500/sec.
Maximum number of metadata columns per GetRecommendations or GetPersonalizedRanking request.	10
Maximum number of recommendation results for a GetRecommendation request without metadata.	500
Maximum number of recommendation results for a GetRecommendation request with metadata.	50
Maximum number of items for ranking in a GetPersonalizedRanking request without metadata.	500
Maximum number of items for ranking in a GetPersonalizedRanking request with metadata.	50
Metric attribution quotas
Maximum number of metrics for a metric attribution	10
Maximum number of unique event attribution sources	100
Batch inference jobs
Maximum number of input files for a batch inference job.	1000
Maximum size of batch inference job input.	1 GB
Maximum number of records per input file for a batch inference job without themes.	50 million
Maximum number of records per input file for a batch inference job with themes.	100
Batch segment jobs
Maximum number of input files for a batch segment job.	1000
Maximum size of batch segment job input.	1 GB
Maximum number of queries per input file for Item-Affinity recipe.	500
Maximum number of queries per input file for Item-Attribute-Affinity recipe.	10
Maximum number of users per segment	5 million
Data deletion jobs
Maximum number of data deletion jobs for a dataset group with a status of PENDING.	5 (adjustable)
Maximum total size of your data deletion input file or files	100 MB

Your AWS account has the following quotas for each region.

Resource	Quota
Total number of active schemas.	500
Total number of active dataset groups.	5 (adjustable)
Total number of pending or in progress dataset import jobs.	5
Total number of pending or in progress batch inference jobs.	5 (adjustable)
Total number of pending or in progress batch segment jobs.	5
Total number of pending or in progress solution versions.	20 (adjustable)

Each dataset group has the following quotas.

Resource	Quota
Total number of active solutions.	10 (adjustable)
Total number of active campaigns.	5 (adjustable)
Total number of recommenders.	5
Total number of filters.	30 (adjustable)
Total number of distinct dataset fields across all filters.	20
Total number of data deletion jobs for a dataset group with a status of PENDING.	5

Requesting a quota increase

For adjustable quotas, you can request a quota increase using the Service Quotas console. The following Amazon Personalize quotas are adjustable:

• Maximum number of item interactions that are considered by a model during training.

• Active campaigns per dataset group

• Active dataset groups

• Active filters per dataset group

• Active solutions per dataset group

• Amount of data per incremental import

• Maximum number of item interactions per user per event type considered by a filter

• Total number of pending or in progress batch inference jobs

• Total number of pending or in progress solution versions

• Maximum rate of PutEvents or PutActionInteraction requests

To request a quota increase, use the Service Quotas console and follow the steps in the Requesting a quota increase section of the Service Quotas User Guide.