@azure/core-xml

Code | Docs

AI Model Inference

@azure-rest/ai-inference

1.0.0 (2026-08-12)

Features Added

• Added secure, cross-platform executable resolution and process-launching primitives for Node.js. #39279

1.0.0 (2026-06-12)

Features Added

This is the first stable version with the package of @azure/arm-weightsandbiases.

1.0.0 (2026-08-01)

This release migrates the package to the @azure/ai-translation-document name and a new client design. It replaces the previous @azure-rest/ai-translation-document REST-level client (RLC) package.

Features Added

• Added support for the 2026-03-01 service API version.
• Added deploymentName to TargetInput and DocumentStatus to support translating with a custom translation model.
• Added deploymentName parameter to single document translation.
• Added support for translating text within images: translateTextWithinImage on batch and single document translation, and image-scan fields (imageCharacterDetected, imageCharged, totalImageScansSucceeded, totalImageScansFailed) on DocumentStatus.

Breaking Changes

• The package name changed from @azure-rest/ai-translation-document to @azure/ai-translation-document.
• The REST-level client (RLC) surface (client.path(...).post(...)) has been replaced with a modeled client surface exposing DocumentTranslationClient and SingleDocumentTranslationClient. See the README for migration examples.

Migration guide

• Install the new package: replace @azure-rest/ai-translation-document with @azure/ai-translation-document.

• Create the client: the default export factory is replaced by named clients.

  - import DocumentTranslator from "@azure-rest/ai-translation-document";
  - const client = DocumentTranslator(endpoint, credential);
  + import { DocumentTranslationClient } from "@azure/ai-translation-document";
  + const client = new DocumentTranslationClient(endpoint, credential);
  

• Call operations: replace the path(...).method() pattern with typed methods.

  - await client.path("/document/batches").post({ body: { inputs } });
  + await client.startTranslation({ inputs });
  

• Single document translation: use SingleDocumentTranslationClient.

  - import DocumentTranslator from "@azure-rest/ai-translation-document";
  - const client = DocumentTranslator(endpoint, credential);
  - await client.path("/document:translate").post({ ... });
  + import { SingleDocumentTranslationClient } from "@azure/ai-translation-document";
  + const client = new SingleDocumentTranslationClient(endpoint, credential);
  + await client.translate(targetLanguage, content);
  

• Responses: results are returned as typed objects and errors are thrown as RestError, instead of checking response.status on a raw HTTP response.

1.0.0 (2026-08-04)

Features Added

• Initial stable release of the Azure AI Discovery client library for JavaScript, exposing WorkspaceClient (conversations, investigations, tasks, tools) and BookshelfClient (knowledge bases).

Code | Docs

Azure AI Search

@azure/search-documents