Google Cloud Translation API
Google Cloud Translation provides programmatic text and document translation with NMT custom models and LLM based adaptive options plus glossary control and enterprise governance under clear per character pricing.
Overview
Cloud Translation supports two editions called Basic and Advanced. Basic offers NMT for text and language detection with a monthly free credit applied across editions. Advanced adds document translation glossary and custom models and now includes LLM based modes for adaptive and text translation.
Pricing is transparent and metered per character with a first half million characters free each month as credit then standard NMT at a per million character rate while LLM adaptive and output metering are documented separately. Teams integrate through REST or client libraries and batch jobs handle large workloads with per page pricing for supported document formats. Glossaries maintain brand terms while datasets can train custom models for domain language.
Security follows Google Cloud policies with regional data handling IAM controls and audit logs. Typical uses include product localization support knowledge translation multilingual chat and analytics. The value is proven scale predictable billing and multiple model paths when cost or quality tradeoffs matter.
Key features
- Two editions called Basic and Advanced so teams pick simple NMT or advanced features like glossary documents and custom training as needs grow
- Transparent metering per character with a monthly free credit so budgets scale with volume and small apps start safely at low cost
- LLM text mode that charges input and output separately so teams trade quality and control while keeping costs predictable by channel
- Adaptive translation that tunes with feedback so recurring content improves over time without a full custom training pipeline
- Glossary support for brand terms and product names so localized content stays consistent across apps pages and support replies
- Document translation for DOCX PPT and PDF by page so batch pipelines scale without manual copy paste and formatting loss
- Client libraries and REST for many languages so integration fits existing stacks and deployment patterns in web and backend jobs
- Enterprise governance with IAM regions and audit logs so regulated teams comply while shipping reliable multilingual features
Best for
- Product localization for apps and docs where glossaries protect terminology and batch jobs process releases with predictable spend
- Support and help center translation that keeps articles current across languages without manual export and external vendor delays
- User generated content moderation and translation where small texts flow through APIs so communities interact across languages
- Multilingual chat and assistants where LLM text mode improves context handling while budgets stay under control for scale
- Analytics and reviews ingestion where snippets get translated at import so teams search and classify content in one language
- Workflow tools that need quick labels and descriptions translated so tickets and issues move without copy paste or external steps
- Education and civic services that publish documents across languages on schedules with audit logs for transparency requirements
- Ecommerce catalogs where attributes and variants localize in bulk while glossaries maintain names specs and compliance details
Capabilities
NMT and LLM Text
Use NMT for cost effective text or LLM text for harder phrasing with separate input and output metering so budgets fit quality needs at scale.
Glossary and Documents
Apply glossaries for brand terms and run DOCX PPT and PDF by page to keep structure while moving large projects through pipelines.
Custom Models
Train domain models when needed and combine with batch jobs so recurring content improves without rebuilding app logic each cycle.
IAM and Regions
Control access logs and data location under Google Cloud policies which supports regulated rollouts and enterprise procurement needs.
Frequently Asked Questions
What is the current entry pricing and free usage?
Cloud Translation applies a monthly free credit worth five hundred thousand characters across editions then charges per million characters as documented on the pricing page.
How do LLM and NMT pricing differ today?
LLM text charges both input and output while standard NMT charges on input only and adaptive LLM has a higher per million character rate for both directions.
Can we translate documents without losing layout?
The Advanced edition supports supported formats with per page pricing that preserves structure better than ad hoc copy and paste routines.
How do glossaries keep brand terminology consistent?
Glossaries map source to approved target terms and apply during calls so names and phrases remain correct across channels and releases.
Is language detection billed separately?
For NMT text calls detection is included and charges only the provided characters so there is no extra fee just for detecting language.
What about data security and residency controls?
Google Cloud IAM audit logging and regional options apply which allows governance choices for regulated industries and public sector teams.
When should we train a custom model vs use LLM text?
Use custom models for narrow domains with stable terminology while LLM text can help with nuanced phrasing and mixed contexts without training.
How can we estimate costs for large imports?
Use the pricing calculator and batch jobs with page pricing to project spend then monitor character counts per release to avoid surprises.
