Skip to main content

Release Notes

2.17.0 - October 22, 2026​

Compatibility Notes​

  • [orchestration, prompt-registry] LlamaGuard38B and LlamaGuard38BFilterConfig are now marked as deprecated and will be removed on 2027-09-20. Migrate usages before that date.

New Features​

  • [context-registry] Add tabular context registry package.
  • [core] Add gemini-3.8-flash to the model types.
  • [core] Add gpt-6-astra, gpt-6-luna, gpt-6.1-sol, gemini-3.6-flash, and nvidia--llama-nemotron-embed-1b to the model types.
  • [langchain] Add reasoning content support to the Orchestration LangChain client. Reasoning blocks from model responses are surfaced in AIMessage.additional_kwargs.reasoning_content(preserving the signature for encrypted thinking), for both non-streaming and streaming responses. On subsequent turns, reasoning is forwarded back to the Orchestration service in reasoning_content from additional_kwargs.reasoning_content. Reasoning placed directly in message content is treated as output only and is not forwarded.
  • [orchestration] Update generated client to latest spec (rel-1.13.2). Add optional provider_specific_fields property to LlmChoice and LlmChoiceStreaming response types.
  • [orchestration] Update the generated client to spec v1.23. Adds the IBM Granite Guardian 4.1 content filter. Adds input filter target selectors. Adds the drop_unsupported_params option on LLMModelDetails.
  • [prompt-registry] Update generated client to latest spec. Add deleteOrchestrationConfigsByScenario, deleteOrchestrationConfigHistory, deletePromptTemplatesByScenario, and deletePromptTemplateHistory API functions.
  • [tabular-orchestration] Add experimental Tabular AI Orchestration package with TabularOrchestrationClient for predictions with Tabular Foundation Models.

Fixed Issues​

  • [langchain] Additional additional_kwargs properties (including tool_calls, intermediate_results, and reasoning_content) are now correctly placed on the AIMessage instead of the generation object.
  • [langchain] Avoid retrying irrecoverable orchestration request and unsupported-model errors.
  • [orchestration] Preserve prompt_tokens_details and completion_tokens_details in streamed token usage to fix missing cache and reasoning token details in OrchestrationStreamResponse.getTokenUsage() and getIntermediateResults().
  • [orchestration] Fix missing prompt template or messages not being detected when no prompt is configured.

Improvements​

  • [core] Remove deprecated models cohere--command-a-reasoning (retirement date: 2026-10-30), mistralai--mistral-medium-instruct (retirement date: 2026-11-30) — use mistralai--mistral-medium instead, gemini-2.5-flash-lite (retirement date: 2026-10-20) — use gemini-3.5-flash-lite instead, sap-rpt-1.5 (retirement date: 2026-12-31) — use sap-rpt-1.6 instead, sap-rpt-1.5-large (retirement date: 2026-12-31) — use sap-rpt-1.6-large instead, sap-rpt-1-small (retirement date: 2026-12-31) — use sap-rpt-1.6 instead and sap-rpt-1-large (retirement date: 2026-12-31) — use sap-rpt-1.6-large instead.

2.16.0 - September 18, 2026​

Compatibility Notes​

  • [openai] Raise the minimum required openai package version to 7.5.0. (875e285)
  • [rpt] Deprecate sap-rpt-1-small and sap-rpt-1-large. These model names remain functional until their retirement date (2026-12-31). (ba5b1b4)
  • [rpt] The server default changed from for parse_data_types changed from true to false. Pass parse_data_types: true explicitly if you relied on the old default. (ba5b1b4)
  • [rpt] The RptClient constructor no longer has a default model name. Pass one of the known model names explicitly, e.g. 'sap-rpt-1.5'. (ba5b1b4)
  • [rpt] Widen TargetColumnConfig.prediction_placeholder to accept null. The type is now string | number | null (was string | number). (ba5b1b4)

New Features​

  • [core, llm-batch] Add LlmBatchModel type for models supported in LLM batch processing. BatchCreateRequest.spec.model now uses this type for IDE autocomplete. (0da56e2)
  • [core] Add gemini-3.5-flash-lite to the available model list. Remove deprecated model mistralai--mistral-large-instruct (retirement date: 2026-09-30). (4b2c014)
  • [rpt] Add a confidence_interval field to PredictResponsePayload predictions for regression tasks. (ba5b1b4)
  • [rpt] Default gzip compression level to 1 for RPT predict requests. (ba5b1b4)
  • [rpt] Add an explanations field to prediction_config; read feature importance scores per query row from the response. (ba5b1b4)
  • [rpt] Expand ColumnType from 3 to 16 values (integer, int16, int32, int64, uint8, decimal, double, boolean, largestring, uuid, time, datetime, timestamp). Map the new numeric variants to number; all other new variants (including datetime and timestamp, which are full ISO strings) map to string. (ba5b1b4)
  • [rpt] Add context_mode to PredictionConfig and PredictResponseMetadata (from RPT spec v1.6.0). (21556db)

Fixed Issues​

  • [core] Set the default http agent socket timeout for AI Core requests to 1201 seconds (1201000 ms). This overrides the Cloud SDK default of 5 seconds, which is too short for chat and streaming completions. The value covers the orchestration server-side maximum of 1200 seconds plus 1 second of leeway. Keep-alive is now disabled on these requests. This avoids reusing stale sockets that a load balancer may have closed during the longer timeout. Both settings apply only to the service-binding destination. You can still override them with a custom destination's agentOptions or CustomRequestConfig.httpsAgent. (3483acb)
  • [langchain] Normalize invalid tool message content to empty string to prevent AI Core 400 errors when MCP tools return empty results. (4cda527)

Improvements​

  • [core] Added mistralai--mistral-medium, sap-rpt-1.5, sap-rpt-1.5-large, sap-rpt-1.6, sap-rpt-1.6-large, gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna to the available model list. Remove deprecated models gpt-4.1-nano (retirement date: 2026-10-14), gemini-2.5-flash (retirement date: 2026-10-16) and gemini-2.5-pro (retirement date: 2026-10-16). (94b36e5)
  • [document-grounding] Update document grounding specification. (478fc53)

2.15.0 - August 18, 2026​

Compatibility Notes​

  • [orchestration] Deprecate OrchestrationConfigRef — use OrchestrationConfigRefById or OrchestrationConfigRefByName instead. (b61cb53)

New Features​

  • [core, openai] Add experimental OpenAI Realtime API (speech-to-speech) support for gpt-realtime via the @sap-ai-sdk/openai/realtime sub-path export. (a5407c8)
  • [langchain] Add orchestrationPromptCachingMiddleware() middleware for the LangChain Orchestration client, exported from @sap-ai-sdk/langchain/orchestration/prompt-caching-middleware. It enables automatic cache control for orchestration requests. The middleware requires the optional langchain peer dependency to be installed. (070b309)
  • [orchestration] Split OrchestrationConfigRef into OrchestrationConfigRefById and OrchestrationConfigRefByName. (b61cb53)

2.14.0 - August 06, 2026​

New Features​

  • [core, orchestration] Add GcpVertexAiEmbeddingModel type with gemini-embedding to @sap-ai-sdk/core. Add GcpVertexAiEmbeddingModel to the EmbeddingModel union in @sap-ai-sdk/orchestration. (cca7edd)
  • [orchestration] Add reasoning content support to the Orchestration client. reasoning_content fields were added on response, streaming delta, and assistant message types. The getReasoningContent() and getDeltaReasoningContent() convenience functions return the reasoning text from model responses. (b986b70)
  • [orchestration] Added overrideConfig field to OrchestrationConfigRef to pass a PartialOrchestrationConfig that overrides parts of the stored orchestration configuration at request time. Streaming via .stream() now automatically sets stream.enabled = true in the partial configuration override, so clients using a stored orchestration configuration reference no longer require streaming to be pre-configured in the stored configuration. (b986b70)
  • [prompt-registry] prompt-registry: Update generated client to latest specification. (a268597)

Fixed Issues​

  • [langchain] Skip tool and assistant messages when applying cache_control, using the last applicable message (system, user, or developer) instead. Some models do not support cache_control on tool messages. (4003a12)
  • [openai] Fix OpenAI client types to allow passing objects with model deployment details (model name + model version, deployment ID and/or resource group) instead of just model names. (016edcd)

Improvements​

  • [core] Add anthropic--claude-4.5-opus, anthropic--claude-4.8-opus, gemini-3.5-flash, gemini-embedding to the available model list. (cca7edd)

2.13.0 - July 14, 2026​

New Features​

  • [core] Add toReadableStream() method to SseStream. The new method converts the asynchronous iterable stream to a pull-based ReadableStream of newline-delimited JSON, enabling composable transformations via the Web Streams API pipeThrough() pattern with native flow control. (6542a2a)
  • [langchain] Add cache_control call option to the LangChain orchestration client. When the cache_control option is set, a cache breakpoint is automatically applied to the request. (fafd99e)
  • [langchain] Expose cached_tokens and cache_creation_tokens in usage_metadata.input_token_details for LangChain orchestration responses. (fafd99e)
  • [openai] Allow setting a model configuration per request. (fea1c3f)

Improvements​

  • [core] Remove deprecated models mistralai--mistral-small-instruct (retirement date: not earlier than 2026-09-30) — use mistralai--mistral-small instead, amazon--nova-premier (retirement date: 2026-09-10) — use amazon--nova-lite (version: 2) instead, anthropic--claude-3-haiku (retirement date: 2026-09-10) — use anthropic--claude-4.5-haiku instead and anthropic--claude-4-sonnet (retirement date: 2026-10-14) — use anthropic--claude-4.5-sonnet instead. (11baf52)
  • [core] Added gpt-5.1 to the available model list. (cb8ffe7)
  • [core] Remove deprecated models gpt-4.1 (retirement date: 2026-10-14), gpt-4.1-mini (retirement date: 2026-10-14), o3 (retirement date: 2026-10-16) and o4-mini (retirement date: 2026-10-16). (a8c9b29)

2.12.0 - June 24, 2026​

New Features​

  • [core] Add AzureOpenAiResponsesModel type for Azure OpenAI models that exclusively support the Responses API. (0f10482)
  • [document-grounding, prompt-registry] Update document grounding specification. (d984914)
  • [foundation-models] Add createBatchInput() and parseBatchOutput() helpers for building and parsing LLM batch requests and responses. (0b41bea)
  • [llm-batch] Add new experimental @sap-ai-sdk/llm-batch package for asynchronous batch processing of LLM requests via SAP AI Core LLM Batch Service. (0b41bea)
  • [openai] Add new experimental @sap-ai-sdk/openai package with a SAP AI Core-aware OpenAI client. The client wraps the OpenAI SDK and handles authentication and deployment resolution against SAP AI Core. (0f10482)

Fixed Issues​

  • [core] Improve error message when server sends a non-JSON response during streaming. (2faf4a8)
  • [langchain] Fixed duplicate tools accumulating in LangChain orchestration client in some configurations on repeated invocations. (311dc08)

Improvements​

  • [core] Added gpt-5.5 to the available model list. Remove retired model anthropic--claude-4-opus — use anthropic--claude-4.7-opus instead. Remove deprecated model o3-mini (retirement date: 2026-08-02). (e944aa1)

2.11.0 - May 26, 2026​

New Features​

  • [langchain] Support orchestration prompt module fallbacks. (dc2f5b1)
  • [orchestration] Update orchestration specification to v0.131.3 (f72bb58)

Fixed Issues​

  • [orchestration] Route messages to messages_history when using an orchestration config reference or prompt template reference. (cbcefb9)
  • [orchestration] Disallow providing both orchestration config reference and config object at the same time. (7db7237)

Improvements​

  • [core] Add gpt-5.4, gpt-5.4-nano to the available model list. Remove deprecated model o1 (retirement date: 2026-06-18). (c8c0e41)
  • [core] Added mistralai--mistral-small, anthropic--claude-4.7-opus, gemini-3.1-flash-lite to the available model list. (75bb9a9)

2.10.0 - April 19, 2026​

Compatibility Notes​

  • [document-grounding] DocumentKeyValueListPair, RetrievalDocumentKeyValueListPair, VectorDocumentKeyValueListPair: the matchMode property type was narrowed from an open union ('ANY' | 'ALL' | any) to the strict FilterMatchModeEnum ('ANY' | 'ALL'). (029f091)
  • [document-grounding] CollectionPendingResponse: fields Location and status were removed. A new monitorURL property was added instead. (029f091)
  • [document-grounding] GoogleDrivePipelineCreateRequest: the configuration property is now required (was optional). Code constructing this request without configuration must be updated to provide it. (94546e6)
  • [document-grounding] GoogleDriveFolderDetail type was removed. Code referencing GoogleDriveFolderDetail must be updated to use GoogleDriveResourceDetail instead. (94546e6)
  • [document-grounding] DataRepositoryType, RetrievalSearchSelectOptionEnum, VectorSearchSelectOptionEnum: previously open (| any) unions are now strictly typed. Only the specified string literals are accepted. (029f091)
  • [document-grounding] S3PipelineMinimalResponse, SFTPPipelineMinimalResponse: the configuration property is now optional. (029f091)
  • [document-grounding] GoogleDriveFolder type was removed and replaced by GoogleDriveResourceDetail. The properties id, driveId, and driverType were replaced by resourceType and resourceId. Code referencing GoogleDriveFolder or its properties must be updated to use GoogleDriveResourceDetail with the new property names. (94546e6)
  • [document-grounding] TextOnlyBaseChunk: new required field id: string added and metadata is now optional. (029f091)
  • [document-grounding] GoogleDriveConfig: the folder property was removed and replaced by a new required resourceType: 'SHARED_FOLDER' | 'SHARED_DRIVE' property, plus optional resourceId and includePaths properties. Code constructing GoogleDriveConfig objects must be updated to use the new shape. (94546e6)
  • [document-grounding] BaseDocument / DocumentInput: chunks type changed from TextOnlyBaseChunk[] to TextOnlyBaseChunkCreate[]. The metadata property is now optional. (029f091)
  • [foundation-models] Support for constructing an AzureOpenAiChatCompletionStreamResponse without an HttpResponse has been deprecated, and will be removed in the next major release. Code directly instantiating this class should be updated to provide an HttpResponse object as the first parameter to allow reading from raw HTTP response. (a722171)
  • [prompt-registry] listPromptTemplateHistory(): the includeSpec query parameter has been removed. (dce18df)
  • [prompt-registry] The include_spec parameter is deprecated in favor of includeSpec and resolve_template_ref is deprecated in favor of resolveTemplateRef. (cc9e80c)
  • [prompt-registry] PromptTemplateSubstitutionRequest now requires the inputParams property. (cc9e80c)

New Features​

  • [document-grounding] Update document grounding specification. (029f091)
  • [foundation-models] Add getRequestId() method to AzureOpenAiChatCompletionResponse, AzureOpenAiChatCompletionStreamResponse and AzureOpenAiEmbeddingResponse. The new method retrieves the request ID from the x-aicore-request-id response header, useful for debugging and tracking requests. (a722171)
  • [foundation-models] Add rawResponse property to AzureOpenAiChatCompletionStreamResponse. The new property exposes the raw HTTP response for advanced use cases such as accessing response headers. (a722171)
  • [orchestration] Update orchestration specification to 0.115.19. Multiple embedding output formats are not yet supported in the orchestration embedding client. (cf767a9)
  • [prompt-registry] Update prompt registry specification. (cc9e80c)

Improvements​

  • [core] Remove deprecated models amazon--titan-embed-image and anthropic--claude-4.5-opus. (8cb466a)

2.9.0 - March 20, 2026​

New Features​

  • [orchestration] Add getCitations() method to retrieve source citations from models like Perplexity Sonar. (8de0013)

Improvements​

  • [core] Add gpt-5.2, anthropic--claude-4.5-opus, anthropic--claude-4.6-opus, anthropic--claude-4.6-sonnet and amazon-titan-embed-image to the available model list. Removed deprecated gpt-4o model. (cd3d8ed)
  • [orchestration] Support file input for user messages. File inputs have a type of file and include a file_data field with a URL such as a HTTP URL or a data URL. Local files must be provided as data:MEDIATYPE;base64,DATA with a non-empty media type and valid base64 content. Availability of different file types depends on the capabilities of the underlying model and tools. GPT-models do not support file inputs with the orchestration API at this time. (b822da9)
  • [rpt] Align parquet endpoint types with the RPT API types. (08c6137)

2.8.0 - March 03, 2026​

New Features​

  • [orchestration] Support streaming with orchestration prompt module fallback. (3d12d4c)
  • [orchestration] Support orchestration prompt module fallback for non-streaming requests When constructing an OrchestrationClient it is now possible to provide a list of module configurations to support module fallback. (5501e7c)
  • [rpt] Add generic HTTP request configuration support. The predictWithSchema() and predictWithoutSchema() methods now accept an optional customRequest parameter of type RptRequestOptions, allowing configuration of custom HTTP request options such as headers, timeout, and middlewares. (e0ef84c)
  • [rpt] Add predict request compression support. All requests with a body of 1024 bytes or larger will be automatically compressed with gzip by default, unless configured otherwise. Compression configuration is available via the requestCompression property on the RptClientConfig object. (e0ef84c)

2.7.0 - February 16, 2026​

Compatibility Notes​

  • [ai-api] Extract functionality around execution scheduling from the ExecutionApi to ExecutionScheduleApi. (8616d5e)
  • [ai-api] Remove the BckndEvent type. (8616d5e)
  • [ai-api] Split the BckndArgoCDApplicationData into BckndArgoCDApplicationDataResponse and BckndArgoCDApplicationDataRequest. (8616d5e)
  • [core] Remove deprecated and retired models from model list. Remove retired model anthropic--claude-3-sonnet from model list, use suggested replacement anthropic--claude-4.5-sonnet instead. Remove retired model anthropic--claude-3-opus from model list. Remove deprecated model gpt-4o-mini, use suggested replacement gpt-5-mini instead. Remove deprecated models anthropic--claude-3.5-sonnet and anthropic--claude-3.7-sonnet. (b12626b)
  • [orchestration] Support for constructing an OrchestrationStreamResponse without an HttpResponse has been deprecated, and will be removed in the next major release. Code directly instantiating this class should be updated to provide an HttpResponse object as the first parameter to allow reading from raw HTTP response. (6b49479)

New Features​

  • [core] Advertise AbortSignal support for HTTP request cancellation. This change adds typings, documentation and examples for using AbortSignal with the HTTP client to enable request cancellation. (b11b00c)
  • [langchain] Support the withStructuredOutput() method in the Orchestration LangChain client. (4fce347)
  • [orchestration] Add rawResponse property to OrchestrationStreamResponse. The new property exposes the raw HTTP response from the orchestration service for advanced use cases. (6b49479)
  • [orchestration] Add getRequestId() method to OrchestrationResponse, OrchestrationStreamResponse and OrchestrationEmbeddingResponse. The new method allows retrieving the request ID from the orchestration service responses, which can be useful for debugging and tracking requests. (6b49479)

Improvements​

  • [core] Combine 'ai-client-type' headers if a custom 'ai-client-type' header is set. (56e9c3f)
  • [rpt] Restrict the task_type property in the PredictionConfig type. (12b4129)

2.6.0 - February 04, 2026​

New Features​

  • [rpt] Release Beta version of a client for the SAP-RPT-1 model. (790ad05)

Fixed Issues​

  • [orchestration] Export isConfigReference() method as a value instead of type-only export. This function can now be imported and used at runtime to check if a configuration is an orchestration configuration reference type. (f3b6dc5)

2.5.0 - January 13, 2026​

Compatibility Notes​

  • [langchain] The @langchain/core package is now a peer dependency. You now have to install the @langchain/core package as a direct dependency of your project. (5b88f6f)

New Features​

  • [langchain] Support disabling streaming completely via the langchain option disableStreaming. (b91e0a7)
  • [langchain] Support auto-streaming via the langchain option streaming. When enabled (e.g., transparently by LangGraph), responses are automatically streamed in invoke() calls. (b91e0a7)
  • [orchestration] Add support for orchestration configuration references in the orchestration client. (2f19a40)

Fixed Issues​

  • [langchain] Pin @langchain/core to v1.1.8 to avoid a regression. (9ff7cec)

Improvements​

  • [langchain] Move the @langchain/core package from dependencies to peer dependencies. This reduces the chance of version incompatibilities between langchain-related packages. (5b88f6f)

2.4.0 - December 23, 2025​

Compatibility Notes​

  • [prompt-registry] In the prompt-registry client schema, the Template type was renamed to PromptTemplate. (a3cbc6e)

New Features​

  • [orchestration] Added support for applyTo and translateMessagesHistory in order to enable selective input translation and automatic inference of target language for output translation parameters. (aa097da)
  • [prompt-registry] Update prompt-registry specification adding support for resource group scoped prompt templates. (a3cbc6e)

Improvements​

  • [core] Added anthropic--claude-4.5-sonnet , anthropic--claude-4.5-haiku , gemini-2.5-flash-lite and sap-abap-1 to the available model list. Removed deprecated models gemini-2.0-flash and gemini-2.0-flash-lite scheduled for retirement. (2e1d2c2)

2.3.0 - November 27, 2025​

Compatibility Notes​

  • [langchain, orchestration, prompt-registry] zod was upgraded to v4 (7c9605d)

New Features​

  • [langchain] Bump langchain to v1 (7c9605d)
  • [orchestration] Added support for protected_material_code property to buildAzureContentSafetyFilter() function for output filter configuration to allow detecting protected code content from known github repositories. (485e21b)

2.2.0 - November 17, 2025​

Compatibility Notes​

  • [document-grounding] MSSharePointConfigurationGetResponse now requires the sharePoint property (e2c34f3)
  • [document-grounding] CommonConfiguration was replaced with backend-specific types: SFTPConfiguration and S3Configuration (e2c34f3)
  • [document-grounding] Some types have been renamed to include endpoint-specific prefixes. Some instances of the prior names may still exist:
    • SearchResults was renamed to VectorSearchResults / RetrievalSearchResults
    • Chunk was renamed to VectorChunk / RetrievalChunk
    • SearchFilter was renamed to VectorSearchFilter
    • KeyValueListPair was renamed to VectorKeyValueListPair / RetrievalKeyValueListPair
    • DocumentKeyValueListPair was renamed to VectorDocumentKeyValueListPair / RetrievalDocumentKeyValueListPair
    • SearchConfiguration was renamed to VectorSearchConfiguration / RetrievalSearchConfiguration
    • SearchSelectOptionEnum was renamed to VectorSearchSelectOptionEnum / RetrievalSearchSelectOptionEnum
    • PerFilterSearchResult was renamed to RetrievalPerFilterSearchResult
    • PerFilterSearchError was renamed to RetrievalPerFilterSearchError
    • DataRepositorySearchResult was renamed to RetrievalDataRepositorySearchResult
    • SearchInput was renamed to RetrievalSearchInput (e2c34f3)
  • [langchain] Remove structured ouput handling for deprecated gpt-4 & gpt-3 models (6100bca)

New Features​

  • [ai-api] Update ai-api package with the new specification (2509b). (58464e9)
  • [core, orchestration] Introduce orchestration embedding client for consuming embedding feature of the orchestration service. (347eac1)
  • [document-grounding] Update document-grounding specification (e2c34f3)

Fixed Issues​

  • [core] Replace active logging during streaming with error throwing to avoid logging the response payload. (5225275)

Improvements​

  • [core, orchestration] Add cohere--command-a-reasoning, mistralai--mistral-medium-instruct and perplexity-ai sonar and sonar-pro to model list (6100bca)

2.1.0 - October 17, 2025​

Fixed Issues​

  • [orchestration] Fix JSDoc example of buildAzureContentSafetyFilter() function. (0cf7d80)

Improvements​

  • [core] Remove alephalpha-pharia-1-7b-control and deepseek-ai--deepseek-r1 from available model list. (0cf7d80)

2.0.0 - September 22, 2025​

Compatibility Notes​

  • [foundation-models, orchestration] Change stream method parameter from AbortController to AbortSignal. The stream() method now accepts an AbortSignal instead of an AbortController as the second parameter in both Azure OpenAI and Orchestration clients. (4c00c27)
  • [foundation-models, orchestration] Response object data property is renamed to _data. Use getter methods like getContent(), getTokenUsage(), getAssistantMessage() instead of direct data access. (5c52cb6)
  • [foundation-models] Move generated types to internal exports while keeping frequently used types in main exports.
    • Generated types are no longer exported from @sap-ai-sdk/foundation-models and must be imported from @sap-ai-sdk/foundation-models/internal.js instead.
    • Frequently used types (AzureOpenAiChatCompletionTool, AzureOpenAiFunctionObject, AzureOpenAiChatCompletionRequestMessage, AzureOpenAiChatCompletionRequestSystemMessage, AzureOpenAiChatCompletionRequestUserMessage, AzureOpenAiChatCompletionRequestAssistantMessage, AzureOpenAiChatCompletionRequestToolMessage) remain available from main package exports.
    • Add new type AzureOpenAiChatCompletionParameters to replace AzureOpenAiCreateChatCompletionRequest which is no longer exported publicly. (5c52cb6)
  • [langchain] Major breaking changes for LangChain orchestration v2:
    • Update LangChain orchestration configuration structure to use promptTemplating instead of separate llm and templating properties.
    • Replace llm.model_name with promptTemplating.model.name and llm.model_params with promptTemplating.model.params.
    • The templating.template property is now promptTemplating.prompt.template.
    • Rename inputParams parameter to placeholderValues in LangChain orchestration client methods.
    • Update message response property names from module_results to intermediate_results in additional kwargs. (86e6370)
  • [langchain] Update imports to use new API facade from foundation-models package.
    • Some generated types now need to be imported from @sap-ai-sdk/foundation-models/internal.js instead of @sap-ai-sdk/foundation-models.
    • Update to use new AzureOpenAiChatCompletionParameters type to replace AzureOpenAiCreateChatCompletionRequest which is no longer exported publicly. (5c52cb6)
  • [orchestration] buildTranslationConfig() function now requires type parameter to distinguish between input and output translation configuration. (740ba78)
  • [orchestration] buildLlamaGuardFilter() function has been renamed to buildLlamaGuard38BFilter(). It now requires a type parameter to distinguish between input and output filter configurations, and accepts filter categories as an array. (740ba78)
  • [orchestration] Move generated types to internal exports while keeping frequently used types in main exports.
    • Generated types are no longer exported from @sap-ai-sdk/orchestration and must be imported from @sap-ai-sdk/orchestration/internal.js instead.
    • Frequently used types (ChatMessage, SystemChatMessage, UserChatMessage, AssistantChatMessage, ToolChatMessage, DeveloperChatMessage, ChatCompletionTool, FunctionObject) remain available from main package exports. (5c52cb6)
  • [orchestration] buildAzureContentSafetyFilter() function now requires type parameter to distinguish between input and output filter configuration. (997e8ec)
  • [orchestration] Major breaking changes for orchestration v2:
    • Consolidate llm and templating modules into a single promptTemplating module.
    • The llm.model_name property is now promptTemplating.model.name and llm.model_params is now promptTemplating.model.params.
    • The templating.template property is now promptTemplating.prompt.template.
    • Rename inputParams parameter to placeholderValues in orchestration client methods.
    • Update response property names from orchestration_result to final_result and module_results to intermediate_results.
    • Replace top-level stream property with streamOptions.enabled and update streaming module options from llm to promptTemplating.
    • Update grounding configuration to use placeholders.input and placeholders.output instead of separate input_params and output_param.
    • Update Azure content filter property names to lowercase with underscores: Hate to hate, SelfHarm to self_harm, Sexual to sexual, and Violence to violence.
    • Remove deprecated buildAzureContentFilter() function and use buildAzureContentSafetyFilter() instead. (86e6370)

New Features​

  • [ai-api] Add resolveDeploymentUrl() function to resolve the deployment URL that matches the given criteria. (14745de)
  • [foundation-models] Add getTokenUsage(), getFinishReason(), getContent(), getToolCalls(), getRefusal(), getAssistantMessage(), findChoiceByIndex() methods to Azure OpenAI chat completion response. (5c52cb6)
  • [orchestration] Add prompt_shield property to buildAzureContentSafetyFilter() function for input filter configuration to allow enabling prompt attack detection. (997e8ec)
  • [orchestration] Add getIntermediateResults() method to OrchestrationResponse, OrchestrationStreamResponse, OrchestrationStreamChunkResponse classes for accessing intermediate processing results from orchestration modules. (5c52cb6)
  • [orchestration] Add deploymentId as the optional parameter for OrchestrationClient initialization. (0a418d0)
  • [orchestration] Add findChoiceByIndex() method to find specific choices by index in streaming responses. (5c52cb6)

Improvements​

  • [core] Add gpt-5,gpt-5-mini and gpt-5-nano to and remove gemini-1.5-flash, gemini-1.5-pro and ibm--granite-13b-chat from the available model list. (500c0dd)
  • [core] Add anthropic--claude-4-opus, anthropic--claude-4-sonnet, amazon--nova-premier, gemini-2.5-flash and gemini-2.5-pro to and remove mistralai--mixtral-8x7b-instruct-v01, meta--llama3.1-70b-instruct, nvidia--llama-3.2-nv-embedqa-1b, amazon--titan-embed-text, gpt-4, amazon--titan-text-express and amazon--titan-text-lite from the available model list. (9e1c43a)