Azure SDK for Python (April 2026)
The Azure SDK team is pleased to announce our April 2026 client library releases.
53 packages released this month.
Stable Packages (7)
-
Core - Client - Core
-
Resource Management - Azure Stack HCI
-
Resource Management - Container Registry
-
Resource Management - Container Service
-
Resource Management - Key Vault
-
Resource Management - Nginx
-
Voice Live
Patch Updates (5)
-
AI Evaluation
-
Azure Monitor OpenTelemetry
-
Resource Management - Key Vault
-
Resource Management - Neon Postgres
-
Resource Management - Storage
Beta Packages (34)
-
AI Agent Server - Core
-
Cosmos DB
-
Load Testing
-
OpenTelemetry Exporter
-
Resource Management - API Management
-
Resource Management - Attestation
-
Resource Management - Azure AI Search
-
Resource Management - Billing Benefits
-
Resource Management - Cognitive Services
-
Resource Management - Confluent
-
Resource Management - Container Registry
-
Resource Management - Container Service
-
Resource Management - Container Service Fleet
-
Resource Management - Device Registry
-
Resource Management - Front Door
-
Resource Management - Kubernetesconfiguration-Extensions
-
Resource Management - Kubernetesconfiguration-Extensiontypes
-
Resource Management - Kubernetesconfiguration-Fluxconfigurations
-
Resource Management - Kubernetesconfiguration-Privatelinkscopes
-
Resource Management - Management Groups
-
Resource Management - Maps
-
Resource Management - NetApp Files
-
Resource Management - Network Function
-
Resource Management - Networkcloud
-
Resource Management - Peering
-
Resource Management - Playwright
-
Resource Management - Purview
-
Resource Management - Redis Enterprise
-
Storage - Blobs
-
Storage - Files Data Lake
-
Storage - Files Share
-
Storage - Queues
-
Synapse - Artifacts
-
Voice Live
Release highlights
AI Agent Server - Core 1.0.0b17 Changelog
Other Changes
- Pin langchain-azure-ai[opentelemetry]==1.1.0b1 to mitigate installation issue.
AI Evaluation 1.16.1 Changelog
Features Added
- Agentic evaluators (Groundedness, ToolCallAccuracy, ToolCallSuccess, ToolInputAccuracy, ToolOutputUtilization, ToolSelection) now accept plain string inputs directly, skipping structured parsing when string format is provided.
Bugs Fixed
- Fixed inconsistency where sample data in evaluation result items did not match the generated sample data from corresponding input rows, ensuring proper synchronization between row-level input samples and their associated evaluation output items.
- Fixed indirect jailbreak (XPIA)
ValueErrorwhen targeting models by convertingbinary_pathfile-based context prompts to inline text before invoking the callback target. - Fixed content-filter responses showing raw JSON API payloads in red team results by detecting blocked responses and replacing them with human-readable messages.
- Fixed missing
token_usageon row-level output items for agent targets using the Foundry execution path by extracting usage data from piece labels. - Fixed 7 backwards-compatibility bugs with the
_use_legacy_endpoint=Truefeature flag including metric name mapping, result parsing, conversation evaluation mode, and Foundry scorer integration.
AI Evaluation 1.16.2 Changelog
Bugs Fixed
- Fixed adversarial chat target incorrectly using user’s callback instead of RAI service, causing callback response to appear as user message in red team scan results when using converter strategies (e.g.,
DIFFICULT,Tense). - Fixed
hate_unfairnessattack success rate always reporting 0% due to metric name mapping using legacyhate_fairnessname instead of canonicalhate_unfairness. - Fixed
TypeErrorin multi-turn and crescendo attacks caused by PyRIT 0.11+ renamingsend_prompt_asyncparameter fromprompt_requesttomessage. - Expanded endpoint normalization (
/openai/v1suffix) to all Azure OpenAI host suffixes (including sovereign clouds), fixing 404 errors for classic AOAI endpoints with PyRIT 0.11+. - Added
aml-aca-tokencredential fallback for agent-type seed requests when running locally without ACA managed identity. - Extract RAI scorer token metrics (prompt_tokens, completion_tokens, total_tokens) into Score metadata and propagate through result processing for downstream aggregation.
AI Evaluation 1.16.3 Changelog
Bugs Fixed
- Fixed attack success rate (ASR) always reporting 0% because the sync eval API’s
passedfield indicates task completion, not content safety. Replacedpassed-based logic with score-based threshold comparison matching_evaluation_processor.py. - Fixed partial red team results being discarded when some objectives fail. Previously, if PyRIT raised due to incomplete objectives (e.g., evaluator model refuses to score), all completed results were lost. Now recovers partial results from PyRIT’s memory database.
- Fixed evaluator token metrics (
promptTokens,completionTokens) not persisted in red teaming output items. The sync eval API returns camelCase keys but the extraction code only checked for snake_case, silently dropping all evaluator token usage data.
Features Added
- Added
extra_headerssupport toOpenAIModelConfigurationto allow passing custom HTTP headers.
AI Evaluation 1.16.4 Changelog
Features Added
- Added support for evaluator
propertiespassthrough in AOAI evaluation results. When an evaluator returns apropertiesdict, it is included alongsidescore,label,reason,threshold, andpassedin the result object.
Bugs Fixed
- Fixed stray space in
_eval_metric.valueattribute access in_base_rai_svc_eval.py.
AI Evaluation 1.16.5 Changelog
Bugs Fixed
- Fixed Jinja2 Server-Side Template Injection (SSTI) vulnerability by replacing unsandboxed
jinja2.Templatewithjinja2.sandbox.SandboxedEnvironmentacross all template rendering paths (CWE-1336). - Fixed
sensitive_data_leakagerisk category producing 0% attack success rate (false negatives) in the Foundry execution path. Agent-specific tool context (e.g.,document_client_smode,email_client_smode) was stored inSeedObjective.metadatabut never propagated to the target callback, so the agent could not access the sensitive data it was supposed to leak. Context is now delivered viaprepended_conversationSeedPrompts and extracted from conversation history metadata, enabling the ACA runtime to build FunctionTool injections. - Fixed multi-turn and crescendo red team strategies producing output items identical to their baseline counterparts. The Foundry execution path was writing all strategies’ conversations to a single shared JSONL file, causing each strategy to read all conversations and mislabel them. Now writes per-strategy JSONL files using PyRIT’s scenario result grouping.
Azure Monitor OpenTelemetry 1.8.7 Changelog
Features Added
- Add Browser SDK loader feature (#42904)
Core - Client - Core 1.39.0 Changelog
Breaking Changes
- Changed the previously undocumented
azure_cloudsetting environment variable fromAZURE_CLOUDtoAZURE_SDK_CLOUD_CONF.
Cosmos DB 4.16.0b1 Changelog
Bugs Fixed
- Fixed bug where a compound session token (containing multiple partition tokens) was sent for single-partition feed range queries. See PR 44484
- Fixed regression where
user_agent_overwritekwarg was not cleaned up properly, causingTypeErrorcrash on sync client construction. See PR 45653 - Fixed bug where client-level
read_timeoutconfiguration was not being automatically applied to all queries. See PR 44472
Features Added
- Added support for Query Advisor feature - See PR 45331
- Added
get_response_headers()andget_last_response_headers()methods to theCosmosItemPagedandCosmosAsyncItemPagedobjects returned byquery_items(), allowing access to response headers from query operations. See PR 44593 - Added InferenceRequestTimeout property for HttpTimeout Policy to Reranking API. See 45469
- Added
full_text_score_scopeparameter toquery_items()for controlling BM25 statistics scope in hybrid search queries. Supports “Local” and “Global” (default) scopes. See 45686
Cosmos DB 4.16.0b2 Changelog
Bugs Fixed
- Fixed bug where container-focused requests using name-based addressing did not consistently populate the
x-ms-cosmos-intended-collection-ridheader. See PR 44080
Load Testing 1.2.0b1 Changelog
Features Added
- This release adds models and enums for all the APIs supported by Azure Load Testing
- Support for Notification Rules
- Added methods
create_or_update_notification_rule,get_notification_rule,delete_notification_ruleandlist_notification_rulesinLoadTestAdministrationClientto work with Notification rules. - Support for Trigger
- Added methods
create_or_update_trigger,get_trigger,delete_triggerandlist_triggersinLoadTestAdministrationClientto work with Scheduling test triggers. - Support for Actionable Insights
- Added methods
begin_generate_test_run_insights,update_latest_test_run_insightsandget_latest_test_run_insightsinLoadTestRunClientto work with actionable insights. - Support for generating test plan recommendations
- Added method
begin_generate_test_plan_recommendationsinLoadTestAdministrationClientwhich generates test plan recommendations. - Support for Cloning load test
- Added method
begin_clone_testinLoadTestAdministrationClientto Clone a load test.
OpenTelemetry Exporter 1.0.0b49 Changelog
Features Added
- Added a 256kb custom properties truncation limit on specific GenAI attributes (#45749)
- Add Browser SDK loader SDK Stats feature bit (#42904)
Bugs Fixed
- Exempt specific GenAI attributes values from custom properties truncation (#45596)
- Fix attributes override in Rate Limited Sampler (#45592)
- Add environment variable to disable/enable custom properties truncation (#45479)
- Fix io counters import issue in performance counters (#45286)
- Remove custom properties truncation (#45118)
- Fix fallback logic when connection string is provided only via environment variable (#45252)
OpenTelemetry Exporter 1.0.0b50 Changelog
Bugs Fixed
- Fix duplicate authentication policy in live metrics exporter causing Unauthorized errors for authenticated Application Insights resources (#46024)
- Suppress internal sdkstats HTTP pipeline logs from appearing in user’s logs (#45966)
- Kubernetes pod name takes precedence when populating
cloud_RoleInstance(#45884)
OpenTelemetry Exporter 1.0.0b51 Changelog
Bugs Fixed
- Added credential authentication support for customer sdkstats (#46143)
Resource Management - API Management 6.0.0b1 Changelog
Resource Management - Attestation 2.0.0b2 Changelog
Resource Management - Azure AI Search 10.0.0b1 Changelog
Resource Management - Azure Stack HCI 8.0.0 Changelog
Resource Management - Billing Benefits 1.0.0b2 Changelog
Resource Management - Cognitive Services 15.0.0b1 Changelog
Resource Management - Confluent 3.0.0b1 Changelog
Resource Management - Container Registry 15.0.0 Changelog
Resource Management - Container Registry 15.1.0b1 Changelog
Resource Management - Container Service 41.0.0 Changelog
Resource Management - Container Service 41.1.0b1 Changelog
Resource Management - Container Service Fleet 4.0.0b3 Changelog
Resource Management - Device Registry 1.2.0b1 Changelog
Resource Management - Front Door 2.0.0b1 Changelog
Resource Management - Key Vault 14.0.0 Changelog
Resource Management - Key Vault 14.0.1 Changelog
Resource Management - Kubernetesconfiguration-Extensions 1.0.0b2 Changelog
Resource Management - Kubernetesconfiguration-Extensiontypes 1.0.0b2 Changelog
Resource Management - Kubernetesconfiguration-Fluxconfigurations 1.0.0b2 Changelog
Resource Management - Kubernetesconfiguration-Privatelinkscopes 1.0.0b2 Changelog
Resource Management - Management Groups 2.0.0b1 Changelog
Resource Management - Maps 3.0.0b1 Changelog
Resource Management - Neon Postgres 1.0.1 Changelog
Resource Management - NetApp Files 15.1.0b1 Changelog
Resource Management - Network Function 1.0.0b2 Changelog
Resource Management - Networkcloud 3.0.0b1 Changelog
Resource Management - Nginx 4.0.0 Changelog
Resource Management - Peering 2.0.0b2 Changelog
Resource Management - Playwright 1.1.0b1 Changelog
Resource Management - Purview 1.1.0b2 Changelog
Resource Management - Redis Enterprise 4.0.0b1 Changelog
Resource Management - Storage 24.0.1 Changelog
Storage - Blobs 12.30.0b1 Changelog
Features Added
- Added support for service version 2026-06-06.
- Added support for connection strings and
account_urls to accept URLs with-ipv6and-dualstacksuffixes forBlobServiceClient,ContainerClient, andBlobClient. - Added support for
createpermission inBlobSasPermissionsforstage_block,stage_block_from_url, andcommit_block_list. - Added support for a new
Smartaccess tier toStandardBlobTierused inBlobClient.set_standard_blob_tier, which is optimized to automatically determine the most cost-effective access with no performance impact. When set,BlobProperties.smart_access_tierwill reveal the service’s current access tier choice betweenHot,Cool, andArchive. - Added support for
is_directorykeyword ingenerate_blob_sasthat generates directory-level SAS for blobs.
Storage - Files Data Lake 12.25.0b1 Changelog
Features Added
- Added support for service version 2026-06-06.
- Added support for connection strings and
account_urls to accept URLs with-ipv6and-dualstacksuffixes forDataLakeServiceClient,FileSystemClient,DataLakeDirectoryClient, andDataLakeFileClient. - Added support for
DataLakeDirectoryClientandDataLakeFileClient’sset_tagsandget_tagsAPIs to conditionally set and get tags associated with a directory or file client, respectively.
Storage - Files Share 12.26.0b1 Changelog
Features Added
- Added support for service version 2026-06-06.
- Added support for the keyword
file_property_semanticsinShareClient’screate_directoryandDirectoryClient’screate_directoryAPIs, which specifies permissions to be configured upon directory creation. - Added support for the keyword
datatoFileClient’screate_fileAPI, which specifies the optional initial data to be uploaded (up to 4MB). - Added support for connection strings and
account_urls to accept URLs with-ipv6and-dualstacksuffixes forShareClient,ShareDirectoryClient, andShareFileClient.
Storage - Queues 12.17.0b1 Changelog
Features Added
- Added support for service version 2026-06-06.
- Added support for connection strings and
account_urls to accept URLs with-ipv6and-dualstacksuffixes forQueueServiceClientandQueueClient.
Synapse - Artifacts 0.22.0 Changelog
Features Added
- Model AzureDatabricksDeltaLakeLinkedService has a new parameter workspace_resource_id
- Model HDInsightOnDemandLinkedService has a new parameter cluster_resource_group_auth_type
- Model JiraObjectDataset has a new parameter table
- Model JiraObjectDataset has a new parameter schema_type_properties_schema
- Model LookupActivity has a new parameter treat_decimal_as_string
- Model NetezzaLinkedService has a new parameter server
- Model NetezzaLinkedService has a new parameter port
- Model NetezzaLinkedService has a new parameter uid
- Model NetezzaLinkedService has a new parameter database
- Model NetezzaLinkedService has a new parameter security_level
- Model SalesforceV2Source has a new parameter partition_option
- Model ScriptActivity has a new parameter treat_decimal_as_string
- Model SnowflakeV2LinkedService has a new parameter treat_decimal_as_string
- Model SnowflakeV2LinkedService has a new parameter use_utc_timestamps
- Added new enum type HDInsightOndemandClusterResourceGroupAuthenticationType
- Added new enum type NetezzaSecurityLevelType
Voice Live 1.2.0b5 Changelog
Features Added
- OpenTelemetry Tracing Support: Added
VoiceLiveInstrumentorfor opt-in OpenTelemetry-based tracing of VoiceLive WebSocket connections, following Azure SDK and GenAI semantic conventions (v1.34.0). Instrumentation covers connection lifecycle (connect,close), message send/receive, and captures voice-specific attributes (gen_ai.voice.session_id,gen_ai.voice.event_type). - Enable via
AZURE_EXPERIMENTAL_ENABLE_GENAI_TRACING=trueenvironment variable. - Content recording controlled by
OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT. - Aligned with
azure-ai-agents/azure-ai-projectstracing model. - Enhanced Telemetry Tracking: Added comprehensive session-level and per-message telemetry:
- Session ID: Automatically captured from
session.created/session.updatedevents and set on the parent connect span (gen_ai.voice.session_id). - Audio format/codec: Input and output audio formats extracted from
session.updatesends (gen_ai.voice.input_audio_format,gen_ai.voice.output_audio_format). - First-token latency: Time from
response.createto firstresponse.audio.deltaorresponse.text.delta, recorded asgen_ai.voice.first_token_latency_ms.response.text.deltais used for latency detection only and is not tracked as a normal recv event. - Turn count: Number of completed responses (
response.done) per session (gen_ai.voice.turn_count). - Interruption count: Number of
response.cancelsends per session (gen_ai.voice.interruption_count). - Audio bytes sent/received: Total audio payload bytes transferred
(
gen_ai.voice.audio_bytes_sent,gen_ai.voice.audio_bytes_received). - Message size: WebSocket message size on each send/recv span
(
gen_ai.voice.message_size). - Rate limit / error events: Server
errorandrate_limits.updatedevents recorded as span events with error codes and rate limit details. - Response & Function Call ID Tracking: All recv and send spans now carry correlation IDs for end-to-end tracing across events:
gen_ai.response.id,gen_ai.conversation.id,gen_ai.voice.call_id,gen_ai.voice.item_id,gen_ai.voice.previous_item_id,gen_ai.voice.output_indexextracted from top-level and nested fields on every event span.gen_ai.response.finish_reasonsfromresponse.doneevents (also propagated to the connect span).- Agent v2 Telemetry: Added agent identity and configuration tracking on the connect span:
gen_ai.agent.idandgen_ai.agent.thread_idextracted fromsession.created/session.updatedserver events.gen_ai.agent.versionandgen_ai.agent.project_namefromAgentSessionConfigat connect time.- MCP (Model Context Protocol) Telemetry: Added tracking for MCP tool calls and approval flows:
- Per-event:
gen_ai.voice.mcp.server_label,gen_ai.voice.mcp.tool_name,gen_ai.voice.mcp.approval_request_id,gen_ai.voice.mcp.approveon recv/send spans. - Session-level:
gen_ai.voice.mcp.call_countandgen_ai.voice.mcp.list_tools_countcounters flushed on session close. - Nested item extraction is guarded by event type to prevent forward-compatibility issues.
Voice Live 1.1.0 Changelog
Features Added
- Added support for Agent configuration through the new
AgentConfigmodel - Added
agentfield toResponseSessionmodel to support agent-based conversations - The
AgentConfigmodel includes properties for agent type, name, description, agent_id, and thread_id
Latest Releases
View all the latest versions of Python packages here.
Installation Instructions
To install the latest beta version of the packages, copy and paste the following commands into a terminal:
$> pip install azure-ai-agentserver-core==1.0.0b17
$> pip install azure-ai-evaluation==1.16.1
$> pip install azure-ai-evaluation==1.16.2
$> pip install azure-ai-evaluation==1.16.3
$> pip install azure-ai-evaluation==1.16.4
$> pip install azure-ai-evaluation==1.16.5
$> pip install azure-ai-voicelive==1.2.0b5
$> pip install azure-ai-voicelive==1.1.0
$> pip install azure-core==1.39.0
$> pip install azure-cosmos==4.16.0b1
$> pip install azure-cosmos==4.16.0b2
$> pip install azure-developer-loadtesting==1.2.0b1
$> pip install azure-mgmt-apimanagement==6.0.0b1
$> pip install azure-mgmt-attestation==2.0.0b2
$> pip install azure-mgmt-azurestackhci==8.0.0
$> pip install azure-mgmt-billingbenefits==1.0.0b2
$> pip install azure-mgmt-cognitiveservices==15.0.0b1
$> pip install azure-mgmt-confluent==3.0.0b1
$> pip install azure-mgmt-containerregistry==15.0.0
$> pip install azure-mgmt-containerregistry==15.1.0b1
$> pip install azure-mgmt-containerservice==41.0.0
$> pip install azure-mgmt-containerservice==41.1.0b1
$> pip install azure-mgmt-containerservicefleet==4.0.0b3
$> pip install azure-mgmt-deviceregistry==1.2.0b1
$> pip install azure-mgmt-frontdoor==2.0.0b1
$> pip install azure-mgmt-keyvault==14.0.0
$> pip install azure-mgmt-keyvault==14.0.1
$> pip install azure-mgmt-kubernetesconfiguration-extensions==1.0.0b2
$> pip install azure-mgmt-kubernetesconfiguration-extensiontypes==1.0.0b2
$> pip install azure-mgmt-kubernetesconfiguration-fluxconfigurations==1.0.0b2
$> pip install azure-mgmt-kubernetesconfiguration-privatelinkscopes==1.0.0b2
$> pip install azure-mgmt-managementgroups==2.0.0b1
$> pip install azure-mgmt-maps==3.0.0b1
$> pip install azure-mgmt-neonpostgres==1.0.1
$> pip install azure-mgmt-netapp==15.1.0b1
$> pip install azure-mgmt-networkcloud==3.0.0b1
$> pip install azure-mgmt-networkfunction==1.0.0b2
$> pip install azure-mgmt-nginx==4.0.0
$> pip install azure-mgmt-peering==2.0.0b2
$> pip install azure-mgmt-playwright==1.1.0b1
$> pip install azure-mgmt-purview==1.1.0b2
$> pip install azure-mgmt-redisenterprise==4.0.0b1
$> pip install azure-mgmt-search==10.0.0b1
$> pip install azure-mgmt-storage==24.0.1
$> pip install azure-monitor-opentelemetry==1.8.7
$> pip install azure-monitor-opentelemetry-exporter==1.0.0b49
$> pip install azure-monitor-opentelemetry-exporter==1.0.0b50
$> pip install azure-monitor-opentelemetry-exporter==1.0.0b51
$> pip install azure-storage-blob==12.30.0b1
$> pip install azure-storage-file-datalake==12.25.0b1
$> pip install azure-storage-file-share==12.26.0b1
$> pip install azure-storage-queue==12.17.0b1
$> pip install azure-synapse-artifacts==0.22.0
Feedback
If you have a bug or feature request for one of the libraries, please post an issue to GitHub.