Azure · dargilco · May 18, 2026 · May 18, 2026 · May 18, 2026 · May 18, 2026
@@ -6,15 +6,22 @@
 
 * New Agent tool `FabricIQPreviewTool`.
 * New Agent tool `ToolboxSearchPreviewTool`.
-* New optional string properties `description` and `name` added to Agent tools which did not have them before.
-* New `.beta.datasets` sub-client with data generation job operations: `create_generation_job`, `get_generation_job`, `list_generation_jobs`, `cancel_generation_job`, `delete_generation_job`.
-* New evaluator generation job operations on `.beta.evaluators`: `create_generation_job`, `get_generation_job`, `list_generation_jobs`, `cancel_generation_job`, `delete_generation_job`.
+* Toolboxes operations promoted from beta to GA: access via `project_client.toolboxes` (was `project_client.beta.toolboxes`).
 * New methods on `.beta.agents` sub-client for code-based hosted agents: `update_agent_from_code()`, `create_agent_version_from_code()`, `download_agent_version_code()`, `download_agent_code()`.
+* New agent optimization operations on `.beta.agents`: `create_optimization_job`, `get_optimization_job`, `list_optimization_jobs`, `cancel_optimization_job`, `list_optimization_candidates`.
+* New evaluator suite operations on `.beta.evaluators`: `create_evaluation_suite`, `list_evaluation_suite_versions`, `get_evaluation_suite`, `delete_evaluation_suite`, `create_generation_suite_job`, `get_generation_suite_job`, `list_generation_suite_jobs`, `cancel_generation_suite_job`.
+* New evaluator generation job operations on `.beta.evaluators`: `create_generation_job`, `get_generation_job`, `list_generation_jobs`, `cancel_generation_job`, `delete_generation_job`.
+* New `.beta.models` sub-client with model weights management operations: `list_versions`, `list`, `get`, `delete`, `create_version`, `upload_file`, `get_pending_upload`.
+* New `.beta.routines` sub-client with routine management operations: `create_or_update`, `get`, `enable`, `disable`, `list`, `delete`, `dispatch_async`.
+* New individual memory operations on `.beta.memory_stores`: `get_memory`, `delete_memory`.
+* New `.beta.datasets` sub-client with data generation job operations: `create_generation_job`, `get_generation_job`, `list_generation_jobs`, `cancel_generation_job`, `delete_generation_job`.
 * New read-only property `content_hash` on `CodeConfiguration`, returning the SHA-256 hex digest of the uploaded code zip.
+* New optional string properties `description` and `name` added to Agent tools which did not have them before.
 
 ### Breaking Changes
 
 Breaking changes in beta methods:
+* `.beta.toolboxes` sub-client removed (promoted to GA as `project_client.toolboxes`).
 * Required keyword `isolation_key` removed from `.beta.agents.create_session()` and `.beta.agents.delete_session()` methods.
 * Argument `body` in methods `.beta.evaluation_taxonomies.create()` and `.beta.evaluation_taxonomies.update()` renamed to `taxonomy`.
 * Argument `body` in method `.beta.skills.create_from_package()` renamed to `content`.
@@ -27,15 +34,16 @@ Breaking changes in beta classes:
 * Renamed class `SessionDirectoryListResponse` to `SessionDirectoryListResult`.
 * Renamed class `SessionFileWriteResponse` to `SessionFileWriteResult`.
 * Renamed class `SkillObject` to `SkillDetails`.
-* Removed class `FabricIQPreviewToolParameters`.
-* Removed class `WorkIQPreviewToolParameters`.
+* Renamed class `Target` to `EvaluationTarget`.
+* Renamed class `TargetConfig` to `RedTeamTargetConfig`.
 
 ### Bugs Fixed
 
 * Fixed telemetry instrumentor to correctly call is_recording() as a method on spans, ensuring non-recording spans are properly skipped (e.g., when sampling is configured) ([GitHub issue 46544](https://github.com/Azure/azure-sdk-for-python/issues/46544)).
 
 ### Sample updates
 
+* Updated Toolbox samples to use GA `project_client.toolboxes` instead of `project_client.beta.toolboxes`.
 * Added Hosted Agent creation samples `sample_create_hosted_agent.py` and `sample_create_hosted_agent_async.py`, demonstrating hosted agent version creation and retrieval with `AIProjectClient`.
 * Added Hosted Agent code-upload samples `sample_create_hosted_agent_from_code.py` and `sample_create_hosted_agent_from_code_async.py`, demonstrating uploading a code package (zip) as a new hosted agent version.
 * The Hosted Agent creation sample also demonstrates assigning the hosted agent managed identity the Azure AI User RBAC role on the backing Azure AI account.

@@ -2,5 +2,5 @@
   "AssetsRepo": "Azure/azure-sdk-assets",
   "AssetsRepoPrefixPath": "python",
   "TagPrefix": "python/ai/azure-ai-projects",
-  "Tag": "python/ai/azure-ai-projects_b8c168cd69"
+  "Tag": "python/ai/azure-ai-projects_d5ecfcf953"
 }
@@ -24,6 +24,7 @@
     DeploymentsOperations,
     EvaluationRulesOperations,
     IndexesOperations,
+    ToolboxesOperations,
 )
 
 if TYPE_CHECKING:
@@ -47,6 +48,8 @@ class AIProjectClient:  # pylint: disable=too-many-instance-attributes
     :vartype deployments: azure.ai.projects.operations.DeploymentsOperations
     :ivar indexes: IndexesOperations operations
     :vartype indexes: azure.ai.projects.operations.IndexesOperations
+    :ivar toolboxes: ToolboxesOperations operations
+    :vartype toolboxes: azure.ai.projects.operations.ToolboxesOperations
     :param endpoint: Foundry Project endpoint in the form
      "https://{ai-services-account-name}.services.ai.azure.com/api/projects/{project-name}". If you
      only have one Project in your Foundry Hub, or to target the default Project in your Hub, use
@@ -102,6 +105,7 @@ def __init__(
         self.datasets = DatasetsOperations(self._client, self._config, self._serialize, self._deserialize)
         self.deployments = DeploymentsOperations(self._client, self._config, self._serialize, self._deserialize)
         self.indexes = IndexesOperations(self._client, self._config, self._serialize, self._deserialize)
+        self.toolboxes = ToolboxesOperations(self._client, self._config, self._serialize, self._deserialize)
 
     def send_request(self, request: HttpRequest, *, stream: bool = False, **kwargs: Any) -> HttpResponse:
         """Runs the network request through the client's chained policies.

@@ -6,8 +6,11 @@
 # Changes may cause incorrect behavior and will be lost if the code is regenerated.
 # --------------------------------------------------------------------------
 
-from typing import TYPE_CHECKING, Union
+from typing import Any, TYPE_CHECKING, Union
 
 if TYPE_CHECKING:
     from . import models as _models
 Filters = Union["_models.ComparisonFilter", "_models.CompoundFilter"]
+ToolCallOutputContent = Union[dict[str, Any], str, list[Any]]
+EvalItemContentItem = Union[str, "_models.EvalItemContentItemObject"]
+EvalItemContent = Union["EvalItemContentItem", list["EvalItemContentItem"]]
@@ -24,6 +24,7 @@
     DeploymentsOperations,
     EvaluationRulesOperations,
     IndexesOperations,
+    ToolboxesOperations,
 )
 
 if TYPE_CHECKING:
@@ -47,6 +48,8 @@ class AIProjectClient:  # pylint: disable=too-many-instance-attributes
     :vartype deployments: azure.ai.projects.aio.operations.DeploymentsOperations
     :ivar indexes: IndexesOperations operations
     :vartype indexes: azure.ai.projects.aio.operations.IndexesOperations
+    :ivar toolboxes: ToolboxesOperations operations
+    :vartype toolboxes: azure.ai.projects.aio.operations.ToolboxesOperations
     :param endpoint: Foundry Project endpoint in the form
      "https://{ai-services-account-name}.services.ai.azure.com/api/projects/{project-name}". If you
      only have one Project in your Foundry Hub, or to target the default Project in your Hub, use
@@ -102,6 +105,7 @@ def __init__(
         self.datasets = DatasetsOperations(self._client, self._config, self._serialize, self._deserialize)
         self.deployments = DeploymentsOperations(self._client, self._config, self._serialize, self._deserialize)
         self.indexes = IndexesOperations(self._client, self._config, self._serialize, self._deserialize)
+        self.toolboxes = ToolboxesOperations(self._client, self._config, self._serialize, self._deserialize)
 
     def send_request(
         self, request: HttpRequest, *, stream: bool = False, **kwargs: Any

@@ -19,6 +19,7 @@
 from ._operations import DatasetsOperations  # type: ignore
 from ._operations import DeploymentsOperations  # type: ignore
 from ._operations import IndexesOperations  # type: ignore
+from ._operations import ToolboxesOperations  # type: ignore
 
 from ._patch import __all__ as _patch_all
 from ._patch import *
@@ -32,6 +33,7 @@
     "DatasetsOperations",
     "DeploymentsOperations",
     "IndexesOperations",
+    "ToolboxesOperations",
 ]
 __all__.extend([p for p in _patch_all if p not in __all__])  # pyright: ignore
 _patch_sdk()