Azure Translator

RESTful API at api.cognitive.microsofttranslator.com. POST body accepts an array of text objects; response mirrors input structure with translations array per text. Language detection via autodetect when 'from' param is omitted. 'to' param accepts comma-separated list for multi-target translation in one call — useful for multilingual agents. Dictionary and transliteration are separate endpoints with identical auth pattern. Document translation is async with S3-like output to Azure Blob Storage.

Structured error responses: HTTP status codes with JSON body containing error code, message, and target field. Common errors documented (400001-400035, 401000, 500000). Rate limits enforced per subscription key with Retry-After headers on 429. Azure SLA: 99.9% uptime guarantee. Quota extension available via Azure support. Document translation provides job status polling and detailed error reporting per file segment.

Azure AI Translator delivers neural machine translation for 100+ language pairs via a straightforward REST API. Key agent capabilities: real-time text translation (batch up to 100 strings per request), document translation for Office/PDF files, dictionary lookup for alternative translations, and transliteration between scripts. Custom Translator portal enables domain-specific fine-tuning. Auth via subscription key or Azure AD token. The API is one of the simplest in the Azure AI stack — minimal setup, clean JSON responses, competitive pricing. Confidence is docs-derived.

Two auth modes: subscription key via Ocp-Apim-Subscription-Key header (simplest), or Azure AD bearer token for managed identity. Regional vs. global endpoints; regional endpoints required for custom models. Key rotation via Azure portal. Ocp-Apim-Subscription-Region header required when using global endpoints — a small but documented gotcha. VNet integration and private endpoints available for enterprise isolation.

Microsoft Learn documentation is comprehensive: quickstarts, tutorials, API reference, and SDK guides for Python, C#, JavaScript, and Java. REST examples in cURL and language-specific SDKs. Interactive API testing via Microsoft's API console. Language support matrix is well-maintained. Getting started requires an Azure account and resource creation, but first call is achievable in minutes with the quickstart guide.