v1.109.0

1.109.0 (2025-08-13)

Features

• Add gpu_partition_size to MachineSpec (b753565)
• Add API for Gen AI Evaluation in Tuning (b753565)
• Add direct_memories_source (b753565)
• Add expiration for TTL for Memory and Sessions (b753565)
• Add force_delete field to DeleteRagFile 'preview' API request for Vertex RAG (b753565)
• Add Grounding with Google Maps tool (b753565)
• Add memory related methods to AdkApp (1368f6a)
• Add option EndpointUserId and ModelUserId fields (b753565)
• Add support for CMEK, runtime controls, and PSC-I to Reasoning Engine protos (b753565)
• Add update_mask "embedding_metadata" to MatchingEngineIndex upsert_datapoints() to support embedding metadata update. (0870512)
• Added document_name for vertex ai search as part of retrieved context from grounding chunk (b753565)
• Expose RecommendSpecs api to vertex python SDK for Custom Weights Model deployment (b753565)
• Migrate dedicated endpoint to be enabled by default (b753565)
• Support for runtime controls and PSC-I (77f7b8e)

Documentation

• A comment for field dedicated_endpoint_enabled in message .google.cloud.aiplatform.v1.DeployRequest is changed (b753565)
• A comment for field monitored_resource_labels in message .google.cloud.aiplatform.v1beta1.AutoscalingMetricSpec is changed (b753565)
• Add encryption_spec to ReasoningEngine (b753565)
• Add psc_interface_config, min/max_instances, resource_limits, container_concurrency to ReasoningEngineSpec (b753565)
• Update comment for allowed values for config models (b753565)

v1.108.0

1.108.0 (2025-08-07)

Features

• Add autoscaling_target_pubsub_num_undelivered_messages option in Preview model deployment on Endpoint & Model classes. (52eacce)

Bug Fixes

• GenAI SDK client - Fix typo in error message for optimize_prompt (63e1caa)
• Set the agent_framework when initializing module-based agent engine (6c15762)

v1.107.0

1.107.0 (2025-08-06)

Features

• A new value NVIDIA_GB200 is added to enum AcceleratorType (d682fac)
• Add DeploymentStage for CreateEndpointOperationMetadata and DeployModelOperationMetadata (d682fac)
• Add a FooBar API (d682fac)
• Add autoscaling metrics parameters for PrivateEndpoint class's model deployment API (58880be)
• Add embedding_metadata to google.cloud.aiplatform.v1.Index (d682fac)
• Add enable_datapoint_upsert_logging to google.cloud.aiplatform.v1.DeployedIndex (d682fac)
• Add exclude_domains for grounding with GoogleSearch and EnterpriseWebSearch (d682fac)
• Add FeatureViewDirectWrite API in v1 (d682fac)
• Add field ReasoningEngineSpec.service_account (d682fac)
• Add ray 2.47 unit tests as required checks (5445648)
• Add the VeoTuningSpec (d682fac)
• Added the ability to use the Model Armor service for content sanitization (d682fac)
• Adds DWS and spot VM feature support to custom batch predictions 2.0 (d682fac)
• GenAI SDK client - add zero-shot prompt optimizer: an option to quickly improve provided system instructions or a prompt, or generate new system instructions based on a prompt. (bc2e8f4)
• GenAI SDK client - Agent Engine Session SDK (8f28c40)
• GenAI SDK client(evals) - add visualization support for rubric-based evaluation workflow (299c44c)
• Online Prediction DeployModel API to support custom metrics based autoscaling (d682fac)
• Remove private preview label from Model Armor protos (d682fac)

Documentation

• Update comments for rpc BatchCreatePullRequestComments, ResolvePullRequestComments and UnresolvePullRequestComments (d682fac)

v1.106.0

1.106.0 (2025-07-30)

Features

• Add service_account parameter to AgentEngine class for creation and update (6359168)
• Add service_account to Reasoning Engine public protos (7b6010b)
• Add service_account to Reasoning Engine public protos (7b6010b)
• Add Vertex AI Model Garden deploy SDK unified Model class in Public Preview (78c8fdd)
• Allow adapter_size=32 for supervised tuning (3a776a7)
• Vertex AI Model Garden deploy SDK Support for self-deploy Partner Model (6c72801)

Bug Fixes

• GenAI SDK client - handle os paths for .whl in requirements when deploying to agent engine (75cd064)
• GenAI SDK client - parsing json body of HTTP response to yield JSON (8cfd9ba)
• Use datasets.MultimodalDataset.from_gemini_request_jsonl() without target_table_id. (e92cca2)

v1.105.0

1.105.0 (2025-07-22)

Features

• Add FlexStart option to DeploymentResourcePool.create, Endpoint.deploy, and Model.deploy (preview) (82dd075)
• Add Ray 2.47 support to RoV Bigquery read/write (8e6df42)
• Add Ray 2.47 support to SDK Client Builder (dde560d)
• Add support for managed oss fine tuning service (2672ec3)
• GenAI SDK client(evals) - Add async evaluate_instances method (a52198a)
• Improve PersistentResource exception logging to include cluster name (4b88698)
• Provide docs for using invoke method. (1315df7)
• RAG - add timeout options for create_corpus, update_corpus and update_rag_engine_config for both v1 and v1beta1 apis. (afa5610)
• Update Ray version support to include Ray v2.47" (e0ee94c)
• Vertex AI Model Garden custom model deploy SDK Public Preview (1ceb2e1)

Bug Fixes

• GenAI SDK client - Fix JS variable name conflict in evals visualization for VS Code iPython environment (079b1f9)

v1.104.0

1.104.0 (2025-07-15)

Features

• Add Aggregation Output in EvaluateDataset Get Operation Response (43eee8d)
• Add API for Managed OSS Fine Tuning (43eee8d)
• Add flexstart option to v1beta1 (43eee8d)
• Expose task_unique_name in pipeline task details for pipeline rerun (43eee8d)
• GenAI SDK client - Add support for context specs when creating agent engine instances (8321826)
• GenAI SDK client(evals) - Add Generate Rubrics API config and internal method (6727fb3)
• GenAI SDK client(evals) - add rubric-based evaluation types (df2390e)
• GenAI SDK client(evals) - Add support for rubric-based metrics, and rubric customization eval workflow (36bfda2)
• Some comments changes in machine_resources.proto to v1beta1 (43eee8d)
• Vertex AI Model Garden custom model deploy Public Preview (43eee8d)

Bug Fixes

• GenAI SDK client(evals) - Handle optional pandas dependency in type hints (cee8d8b)

Documentation

• A comment for field boot_disk_type in message .google.cloud.aiplatform.v1beta1.DiskSpec is changed (43eee8d)
• A comment for field learning_rate_multiplier in message .google.cloud.aiplatform.v1beta1.SupervisedHyperParameters is changed (43eee8d)
• A comment for field machine_spec in message .google.cloud.aiplatform.v1beta1.DedicatedResources is changed (43eee8d)
• A comment for field max_replica_count in message .google.cloud.aiplatform.v1beta1.AutomaticResources is changed (43eee8d)
• A comment for field max_replica_count in message .google.cloud.aiplatform.v1beta1.DedicatedResources is changed (43eee8d)
• A comment for field min_replica_count in message .google.cloud.aiplatform.v1beta1.AutomaticResources is changed (43eee8d)
• A comment for field min_replica_count in message .google.cloud.aiplatform.v1beta1.DedicatedResources is changed (43eee8d)
• A comment for field model in message .google.cloud.aiplatform.v1beta1.TunedModel is changed (43eee8d)
• A comment for field required_replica_count in message .google.cloud.aiplatform.v1beta1.DedicatedResources is changed (43eee8d)
• A comment for field training_dataset_uri in message .google.cloud.aiplatform.v1beta1.SupervisedTuningSpec is changed (43eee8d)
• A comment for field validation_dataset_uri in message .google.cloud.aiplatform.v1beta1.SupervisedTuningSpec is changed (43eee8d)
• A comment for message DedicatedResources is changed (43eee8d)
• Add constraints for AggregationMetric enum and default value for flip_enabled field in AutoraterConfig (43eee8d)

v1.103.0

1.103.0 (2025-07-10)

Features

• Add ADK version check and set MemoryBankService as default when google-adk>=1.5.0 (262fbc3)
• Add logging for agent engine creation (795ee17)
• Populate task_unique_name from initial pipeline run in Pipeline Task Rerun Configs for pipeline job rerun (116a0a6)
• Ummd.MultimodalDataset.from_bigquery() now also accepts a table id (not just a BQ table URI). (6e5c421)

v1.102.0

1.102.0 (2025-07-08)

Features

• Add message ColabImage, add field colab_image to NotebookSoftwareConfig (2c64a76)
• Configure Bigframes implicitly in MultimodalDataset.assess(). (0664ea3)
• GenAI SDK client - add async version of prompt optimizer (4564c9c)
• GenAI SDK client (evals) - add LLMMetric.load function to load a config file (local or GCS) (56252e8)

Documentation

• Fix the docstring example for unary Endpoint invoke method. (a132e86)

v1.101.0

1.101.0 (2025-07-01)

Features

• Allow installation scripts in AgentEngine. (9296d4d)
• Add invoke method. It supports both streaming and non-streaming cases. (e686932)
• Add computer use support to tools (f56c42e)
• Add computer use support to tools (f56c42e)
• Allow users to pass project_number for custom job service account when service_account is not provided. (5b59030)
• Expose task_unique_name in pipeline task details for pipeline rerun (f56c42e)
• Support creating an invoke enabled model in Python SDK (71a8d7b)

v1.100.0

1.100.0 (2025-06-26)

Features

• Add import_embeddings method in MatchingEngineIndex resource (5a0df36)
• Add invoke_route_prefix to ModelContainerSpec in aiplatform v1 models.proto (4202177)
• Add invoke_route_prefix to ModelContainerSpec in aiplatform v1beta1 models.proto (d4ede02)
• Add Model Garden deploy OSS model API (d4ede02)
• Add PSCAutomationConfig to PrivateServiceConnectConfig in service_networking.proto (d4ede02)
• Add validation assessment for batch prediction. (d570fc9)
• GenAI SDK client - Add batch_evaluate method for asynchronous batch eval. Add transformation support for consistent interface parameters with the evaluate method (4d44f94)
• GenAI SDK client - Add Vertex AI Prompt Optimizer to the Gen AI SDK (experimental) (5daacda)
• GenAI SDK client - Initial release of Agent Engine Memories SDK (e8d18b6)
• GenAI SDK client (evals) - add support for third-party model inference via litellm library (e728d8b)
• matching-engine: Add sync argument to deploy_index (fee1e2d)
• Reasoning Engine v1beta1 subresource updates (d4ede02)
• Updated explicit sync to existing decorator optional_sync (fee1e2d)

Bug Fixes

• Fix auth scope for RAG upload_file (a506b94)
• Fixed return type for deploy_index and added test for sync values (fee1e2d)
• Use PrivateServiceConnectConfig in service_networking in PrivateEndpoint instead of the wrapper class PrivateServiceConnectConfig (87c2c3e)