AutoML Translation API isdeprecatedand will no longer be available on Google Cloud after September 30, 2025. The functionality and management of custom models is available throughCloud Translate API - Advanced (v3).
Stay organized with collectionsSave and categorize content based on your preferences.
This page documents production updates to AutoML Translation. We recommend
that AutoML Translation developers periodically check this list for any
new announcements. Major changes will also be announced via thegoogle-translate-apimailing list.
You can see the latest product updates for all of Google Cloud on theGoogle Cloudpage, browse and filter all release notes in theGoogle Cloud console,
or programmatically access release notes inBigQuery.
To get the latest product updates delivered to you, add the URL of this page to yourfeed
reader, or add thefeed URLdirectly.
September 16, 2024
AutoML Translation API isdeprecatedand will no longer be available on Google Cloud after September 30, 2025. You can replicate the functionality of custom models throughCloud Translate - Advanced (v3).
AutoML Translation is now generally available. Translations are now faster and quality should be improved or on par with previous versions; individual results will vary depending on your datasets. In addition, quality improvements can vary across language pairs.
Model lifetimeAutoML Models run on a managed platform that is regularly upgraded. AutoML Models need to be re-trained at least every 18 months to ensure consistent performance and stability in translations over time and to get any benefits of the upgrades since original training. Models created before AutoML GA (for example, models created during service Alpha /Beta) need to be re-trained within 18 months of GA to move all assets to the newest version of the platform.
AutoML Translation has upgraded its Google NMT base models onto a new training architecture so that it now runs end-to-end on upgradedTPUs. Going forward, all training using Google NMT as the base model trains and serves on TPUs.
Customers already using AutoML might experience additional differences between v1 and previous versions of the service:
Data validation now provides warnings on data import to
indicate certain data is not usable for training jobs.
Data import takes longer in order to support data validation.
To achieve higher quality on the new
architecture, training durations are longer.
September 04, 2019
AutoML Translation has enabled Google Cloud Platform (GCP) Console-based UI for all its customers. This improves the Cloud AutoML Translation experience by adding UI elements such as controls to sort/filter datasets, which provides consistency with the rest of GCP and allows for faster release of new features.
The URL for the AutoML Translation UI has changed to https://console.cloud.google.com/translation.
December 11, 2018
The minimum number of sentence pairs required to train a custom model has changed. If you don't explicitly specify training, validation, and test sets, the minimum number of sentence pairs is 1000. If you do explicitly specify training, validation, and test sets, the minimum number of pairs used for validation and test is 100 each.
November 01, 2018
The AutoML Translation service account requires additional permissions to access project resources when using the AutoML Translation UI. You must add the roleserviceusage.serviceUsageAdmin. See step 8 inBefore You Begin.
September 18, 2018
AutoML Translation supports many new languages as source or target languages paired with English, bringing the total number of supported language pairs to 100. SeeLanguage Support for Custom Modelsfor the complete list.
August 24, 2018
The URL for the AutoML Translation UI has changed to https://cloud.google.com/automl/ui/translation. The previous URL https://beta-dot-custom-vision.appspot.com/translation/ will remain active.
August 02, 2018
AutoML Translation supports Swahili (sw) as a target language from English (en) as a source language.
Language Pair
Language Codes
Swahili <- English
sw<-en
July 24, 2018
Models created during the EAP will continue to be valid for the Beta. You do not need to recreate/retrain models. However, the format of resource IDs (dataset IDs, model IDs, and operation IDs) is changing. If you are using hardcoded resource IDs, you need to obtain new IDs for those resources (using the get or list method) by August 31, 2018.
May 24, 2018
The refresh of the AutoML Translation EAP release introduces a new UI design and a new API. The previous version of the AutoML Translation EAP UI is deprecated, and the new AutoML API is not compatible with the previous API.
You must reimport the contents any previously created datasets. You can find the original contents atgs://{project-id}-vcm/{dataset-name}/uploads/{document-name}.txt.
If you used the previous version of the AutoML Translation UI to create custom models based on other custom translation models (rather than the default Google Neural Machine Translation model), you will also need to recreate or retrain those models using the new UI.
March 30, 2018
The AutoML Translation EAP web application now enables you to build a custom model based on another custom translation model that you built.
A sample Java application is available that demonstrates how to send a request for translation and obtain the response back.
February 27, 2018
The following language translation pairs have been added:
Language Pair
Language Codes
Dutch -> English
nl->en
Hebrew -> English
iw->en
February 05, 2018
The following language translation pairs have been added:
[[["Easy to understand","easyToUnderstand","thumb-up"],["Solved my problem","solvedMyProblem","thumb-up"],["Other","otherUp","thumb-up"]],[["Hard to understand","hardToUnderstand","thumb-down"],["Incorrect information or sample code","incorrectInformationOrSampleCode","thumb-down"],["Missing the information/samples I need","missingTheInformationSamplesINeed","thumb-down"],["Other","otherDown","thumb-down"]],["Last updated 2025-09-04 UTC."],[],[],null,["# AutoML Translation release notes\n\nThis page documents production updates to AutoML Translation. We recommend that AutoML Translation developers periodically check this list for any new announcements. Major changes will also be announced via the [google-translate-api](https://groups.google.com/forum/#!forum/google-translate-api) mailing list.\n\n\u003cbr /\u003e\n\n\nYou can see the latest product updates for all of Google Cloud on the\n[Google Cloud](/release-notes) page, browse and filter all release notes in the\n[Google Cloud console](https://console.cloud.google.com/release-notes),\nor programmatically access release notes in\n[BigQuery](https://console.cloud.google.com/bigquery?p=bigquery-public-data&d=google_cloud_release_notes&t=release_notes&page=table).\n\nTo get the latest product updates delivered to you, add the URL of this page to your\n[feed\nreader](https://wikipedia.org/wiki/Comparison_of_feed_aggregators), or add the\n[feed URL](https://cloud.google.com/feeds/automl-translate-release-notes.xml) directly.\n\nSeptember 16, 2024\n------------------\n\nAutoML Translation API is [deprecated](https://cloud.google.com/translate/automl/docs/deprecations) and will no longer be available on Google Cloud after September 30, 2025. You can replicate the functionality of custom models through [Cloud Translate - Advanced (v3)](https://cloud.google.com/translate/docs).\n\nOctober 28, 2022\n----------------\n\nThe following [language translation pairs](https://cloud.google.com/translate/automl/docs/languages) have been added:\n\nJuly 17, 2020\n-------------\n\nFor test data, added support for the `.tmx` file type when evaluating existing models. For more information, see [Evaluating models](https://cloud.google.com/translate/automl/docs/evaluate#evaluate_and_compare_models_using_a_new_test_set).\n\nJuly 01, 2020\n-------------\n\nAutoML Translation no longer supports incremental model training. Instead, train new models on datasets that combine existing and incremental data.\n\nApril 03, 2020\n--------------\n\nIntegration with [VPC Service Controls](https://cloud.google.com/vpc-service-controls/docs/overview) is now in [beta stage](https://cloud.google.com/products/#product-launch-stages).\n\nOctober 14, 2019\n----------------\n\nAutoML Translation is now generally available. Translations are now faster and quality should be improved or on par with previous versions; individual results will vary depending on your datasets. In addition, quality improvements can vary across language pairs. \n**Model lifetime** AutoML Models run on a managed platform that is regularly upgraded. AutoML Models need to be re-trained at least every 18 months to ensure consistent performance and stability in translations over time and to get any benefits of the upgrades since original training. Models created before AutoML GA (for example, models created during service Alpha /Beta) need to be re-trained within 18 months of GA to move all assets to the newest version of the platform. \nAutoML Translation has upgraded its Google NMT base models onto a new training architecture so that it now runs end-to-end on upgraded\n[TPUs](https://cloud.google.com/tpu/docs/tpus). Going forward, all training using Google NMT as the base model trains and serves on TPUs. \nCustomers already using AutoML might experience additional differences between v1 and previous versions of the service:\n\n- Data validation now provides warnings on data import to indicate certain data is not usable for training jobs.\n- Data import takes longer in order to support data validation.\n- To achieve higher quality on the new architecture, training durations are longer.\n\nSeptember 04, 2019\n------------------\n\nAutoML Translation has enabled Google Cloud Platform (GCP) Console-based UI for all its customers. This improves the Cloud AutoML Translation experience by adding UI elements such as controls to sort/filter datasets, which provides consistency with the rest of GCP and allows for faster release of new features. \nThe URL for the AutoML Translation UI has changed to https://console.cloud.google.com/translation.\n\nDecember 11, 2018\n-----------------\n\nThe minimum number of sentence pairs required to train a custom model has changed. If you don't explicitly specify training, validation, and test sets, the minimum number of sentence pairs is 1000. If you do explicitly specify training, validation, and test sets, the minimum number of pairs used for validation and test is 100 each.\n\nNovember 01, 2018\n-----------------\n\nThe AutoML Translation service account requires additional permissions to access project resources when using the AutoML Translation UI. You must add the role `serviceusage.serviceUsageAdmin`. See step 8 in [Before You Begin](https://cloud.google.com/translate/automl/docs/before-you-begin).\n\nSeptember 18, 2018\n------------------\n\nAutoML Translation supports many new languages as source or target languages paired with English, bringing the total number of supported language pairs to 100. See [Language Support for Custom Models](https://cloud.google.com/translate/automl/docs/languages) for the complete list.\n\nAugust 24, 2018\n---------------\n\nThe URL for the AutoML Translation UI has changed to https://cloud.google.com/automl/ui/translation. The previous URL https://beta-dot-custom-vision.appspot.com/translation/ will remain active.\n\nAugust 02, 2018\n---------------\n\nAutoML Translation supports Swahili (`sw`) as a target language from English (`en`) as a source language.\n\nJuly 24, 2018\n-------------\n\nModels created during the EAP will continue to be valid for the Beta. You do not need to recreate/retrain models. However, the format of resource IDs (dataset IDs, model IDs, and operation IDs) is changing. If you are using hardcoded resource IDs, you need to obtain new IDs for those resources (using the get or list method) by August 31, 2018.\n| **Note:** Enabling the AutoML API for your project automatically also enables the Cloud Vision API, Cloud Translation API, and Cloud Natural Language API. Disabling any of these three APIs will also disable the AutoML API\n\nMay 24, 2018\n------------\n\nThe refresh of the AutoML Translation EAP release introduces a new UI design and a new API. The previous version of the AutoML Translation EAP UI is deprecated, and the new AutoML API is not compatible with the previous API.\n\nYou must reimport the contents any previously created datasets. You can find the original contents at `gs://{project-id}-vcm/{dataset-name}/uploads/{document-name}.txt`.\n\nIf you used the previous version of the AutoML Translation UI to create custom models based on other custom translation models (rather than the default Google Neural Machine Translation model), you will also need to recreate or retrain those models using the new UI.\n\nMarch 30, 2018\n--------------\n\n- The AutoML Translation EAP web application now enables you to build a custom model based on another custom translation model that you built.\n- A sample Java application is available that demonstrates how to send a request for translation and obtain the response back.\n\nFebruary 27, 2018\n-----------------\n\nThe following language translation pairs have been added:\n\nFebruary 05, 2018\n-----------------\n\nThe following language translation pairs have been added:\n\nJanuary 17, 2018\n----------------\n\nAutoML Translation EAP release."]]
