java release notes for the 2024-09 release (#7942)

_data/releases/2024-09/java.yml

@@ -1224,4 +1224,55 @@ entries:
 
     - Updated Azure SDK dependency versions to the latest releases.
   GroupId: com.azure
+- Name: azure-ai-openai-assistants
+  Version: 1.0.0-beta.4
+  DisplayName: OpenAI Assistants
+  ServiceName: Cognitive Services
+  VersionType: Beta
+  Hidden: false
+  ChangelogUrl: https://github.com/Azure/azure-sdk-for-java/tree/azure-ai-openai-assistants_1.0.0-beta.4/sdk/openai/azure-ai-openai-assistants/CHANGELOG.md
+  ChangelogContent: |-
+    #### Features Added
+
+    - `file_search` tool definitions now have the ability to configure `max_num_results` via the matching, named inner options object
+    Previously, only a stubbed `{ "type": "file_search" }` was valid. Now, e.g.: `{ "type": "file_search", "file_search": { "max_num_results": 20 } }`
+    - Added a new property `FileSearchToolDefinitionDetails fileSearch` to `FileSearchToolDefinition` model.
+    - Added new class `FileSearchToolDefinitionDetails` to represent the details of a file search tool.
+
+    - `chunking_strategy` is added as a new optional property when creating a vector store (either via the vector store creation operation
+    or the helper when creating an assistant) -- this allows customization of the chunk size and overlap used when ingesting data.
+    See the OpenAI reference for full details.
+    - Added a new property `VectorStoreChunkingStrategyRequest chunkingStrategy` to `VectorStoreOptions` model.
+    - Added a new property `VectorStoreChunkingStrategyResponse chunkingStrategy` to `VectorStoreFile` model.
+    - Added new enum `VectorStoreChunkingStrategyRequestType` and `VectorStoreChunkingStrategyResponseType` to represent the chunking strategy `type` for vector stores.
+    - Added new class `VectorStoreChunkingStrategyRequest` and `VectorStoreChunkingStrategyResponse` to represent the chunking strategy for vector stores.
+    - Added new class `VectorStoreAutoChunkingStrategyRequest` and `VectorStoreAutoChunkingStrategyResponse` to represent the `auto` chunking strategy for vector stores.
+    - Added new class `VectorStoreStaticChunkingStrategyOptions`, `VectorStoreStaticChunkingStrategyRequest` and `VectorStoreStaticChunkingStrategyResponse`
+    to represent the `static` chunking strategy for vector stores.
+  GroupId: com.azure
+- Name: azure-ai-openai
+  Version: 1.0.0-beta.11
+  DisplayName: OpenAI
+  ServiceName: Cognitive Services
+  VersionType: Beta
+  Hidden: false
+  ChangelogUrl: https://github.com/Azure/azure-sdk-for-java/tree/azure-ai-openai_1.0.0-beta.11/sdk/openai/azure-ai-openai/CHANGELOG.md
+  ChangelogContent: |-
+    #### Features Added
+
+    - Added a new overload `getImageGenerationsWithResponse` that takes `RequestOptions` to provide the flexibility to
+    modify the HTTP request.
+    - Added the capability to handle the float[] `embedding` type when serializing to `toJson()` and deserializing from `fromJson()` in the `EmbeddingItem` class. ([#41159](https://github.com/Azure/azure-sdk-for-java/issues/41159))
+    - A new `required` keyword is added to `tool_choice` in the request options, which specifies that at least one tool must be called. This adds to the existing `auto` (default), `none`, and ability to specify a specific (function) tool by name
+    - New operation paths: `/batches`, `/batches/{batch-id}` and `/batches/{batch-id}/cancel`
+    - New request model: `BatchCreateRequest`
+    - New response models: `Batch`, `BatchErrors`, etc.
+    - Added `/files` operation from the Azure OpenAI Assistants SDK
+    - New `batch` and `batch_output` files purposes
+    - In `ChatCompletionsOptions` the `setToolChoice` method uses `ChatCompletionsToolSelection` to pass either a preset enum or a named tool.
+
+    #### Breaking Changes
+
+    - Removed `finish_details` field from `ChatChoice` class since service API version, `2024-07-01-preview`, does not return this field.
+  GroupId: com.azure