diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/CHANGELOG.md b/sdk/contentunderstanding/azure-ai-contentunderstanding/CHANGELOG.md
index 948c75550a64..c1ca68646dc0 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/CHANGELOG.md
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/CHANGELOG.md
@@ -1,5 +1,10 @@
 # Release History
 
+## 1.0.2 (2026-04-20)
+
+### Bugs Fixed
+- Exposed `usage` property on `AnalyzeLROPoller` and `AnalyzeAsyncLROPoller` to surface billing and token consumption details (`UsageDetails`) returned by the REST API. Previously the `usage` field in the LRO response envelope was discarded by the generated deserialization callback.
+
 ## 1.0.1 (2026-03-06)
 
 ### Bugs Fixed
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/README.md b/sdk/contentunderstanding/azure-ai-contentunderstanding/README.md
index fd8d79ffbbaa..9cb82881a725 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/README.md
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/README.md
@@ -31,6 +31,7 @@ This table shows the relationship between SDK versions and supported API service
 
 | SDK version | Supported API service version |
 | ----------- | ----------------------------- |
+| 1.0.2       | 2025-11-01                    |
 | 1.0.1       | 2025-11-01                    |
 | 1.0.0       | 2025-11-01                    |
 
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/azure/ai/contentunderstanding/_version.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/azure/ai/contentunderstanding/_version.py
index 83602e6274bc..8e44259d1a18 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/azure/ai/contentunderstanding/_version.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/azure/ai/contentunderstanding/_version.py
@@ -6,4 +6,4 @@
 # Changes may cause incorrect behavior and will be lost if the code is regenerated.
 # --------------------------------------------------------------------------
 
-VERSION = "1.0.1"
+VERSION = "1.0.2"
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/azure/ai/contentunderstanding/aio/models/_patch.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/azure/ai/contentunderstanding/aio/models/_patch.py
index eca66620c5d4..df2ac0fdbe6b 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/azure/ai/contentunderstanding/aio/models/_patch.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/azure/ai/contentunderstanding/aio/models/_patch.py
@@ -10,8 +10,10 @@
 """
 
 import re
-from typing import Any, TypeVar
+from typing import Any, Optional, TypeVar
 from azure.core.polling import AsyncLROPoller, AsyncPollingMethod
+from ... import models as _models
+from ..._utils.model_base import _deserialize
 
 PollingReturnType_co = TypeVar("PollingReturnType_co", covariant=True)
 
@@ -62,6 +64,26 @@ def operation_id(self) -> str:
         except (KeyError, ValueError) as e:
             raise ValueError(f"Could not extract operation ID: {str(e)}") from e
 
+    @property
+    def usage(self) -> Optional["_models.UsageDetails"]:
+        """Returns the usage details from the completed analyze operation.
+
+        This property is available after the operation has completed successfully.
+        It provides information about the resources consumed during analysis,
+        including document pages, contextualization tokens, and LLM token breakdown.
+
+        :return: The usage details, or None if the operation is not yet complete
+            or usage information is not available.
+        :rtype: ~azure.ai.contentunderstanding.models.UsageDetails or None
+        """
+        if not self.done():
+            return None
+        response = self.polling_method()._pipeline_response.http_response  # type: ignore # pylint: disable=protected-access
+        usage_data = response.json().get("usage")
+        if usage_data is None:
+            return None
+        return _deserialize(_models.UsageDetails, usage_data)
+
     @classmethod
     def from_poller(cls, poller: AsyncLROPoller[PollingReturnType_co]) -> "AnalyzeAsyncLROPoller[PollingReturnType_co]":  # pyright: ignore[reportInvalidTypeArguments]  # fmt: skip
         """Wrap an existing AsyncLROPoller without re-initializing the polling method.
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/azure/ai/contentunderstanding/models/_patch.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/azure/ai/contentunderstanding/models/_patch.py
index 9dee69387cba..199ca653e0c7 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/azure/ai/contentunderstanding/models/_patch.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/azure/ai/contentunderstanding/models/_patch.py
@@ -14,6 +14,8 @@
 from typing import TYPE_CHECKING, Any, Dict, List, Optional, TypeVar
 from azure.core import CaseInsensitiveEnumMeta
 from azure.core.polling import LROPoller, PollingMethod
+from . import _models
+from .._utils.model_base import _deserialize
 from ._models import (
     StringField,
     IntegerField,
@@ -155,6 +157,26 @@ def operation_id(self) -> str:
         except (KeyError, ValueError) as e:
             raise ValueError(f"Could not extract operation ID: {str(e)}") from e
 
+    @property
+    def usage(self) -> Optional["_models.UsageDetails"]:
+        """Returns the usage details from the completed analyze operation.
+
+        This property is available after the operation has completed successfully.
+        It provides information about the resources consumed during analysis,
+        including document pages, contextualization tokens, and LLM token breakdown.
+
+        :return: The usage details, or None if the operation is not yet complete
+            or usage information is not available.
+        :rtype: ~azure.ai.contentunderstanding.models.UsageDetails or None
+        """
+        if not self.done():
+            return None
+        response = self.polling_method()._pipeline_response.http_response  # type: ignore # pylint: disable=protected-access
+        usage_data = response.json().get("usage")
+        if usage_data is None:
+            return None
+        return _deserialize(_models.UsageDetails, usage_data)
+
     @classmethod
     def from_poller(cls, poller: LROPoller[PollingReturnType_co]) -> "AnalyzeLROPoller[PollingReturnType_co]":  # pyright: ignore[reportInvalidTypeArguments]  # fmt: skip
         """Wrap an existing LROPoller without re-initializing the polling method.
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/README.md b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/README.md
index d96bc5f0cfc2..70ee18191d3f 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/README.md
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/README.md
@@ -63,6 +63,7 @@ Extracts structured fields from invoices using `prebuilt-invoice` analyzer. Show
 - Extracting structured fields (customer name, totals, dates, line items)
 - Working with field confidence scores and source locations
 - Accessing object fields and array fields
+- Accessing usage details (billing metrics, token consumption per model)
 - Financial document processing (invoices, receipts, credit cards, bank statements, checks)
 
 ### Sample 04: Create Analyzer
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_analyze_binary_async.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_analyze_binary_async.py
index 693c28b719d4..f3c9c9c62403 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_analyze_binary_async.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_analyze_binary_async.py
@@ -107,7 +107,7 @@ async def main() -> None:
         if isinstance(range_result.contents[0], DocumentContent):
             range_doc = range_result.contents[0]
             print(
-                f"Content range analysis returned pages"
+                "Content range analysis returned pages"
                 f" {range_doc.start_page_number} - {range_doc.end_page_number}"
             )
         # [END analyze_binary_with_content_range]
@@ -125,7 +125,7 @@ async def main() -> None:
         if isinstance(combine_range_result.contents[0], DocumentContent):
             combine_doc = combine_range_result.contents[0]
             print(
-                f"Combined content range analysis returned pages"
+                "Combined content range analysis returned pages"
                 f" {combine_doc.start_page_number} - {combine_doc.end_page_number}"
             )
         # [END analyze_binary_with_combined_content_range]
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_analyze_invoice_async.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_analyze_invoice_async.py
index 6cbea029dc7d..62da86d9955c 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_analyze_invoice_async.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_analyze_invoice_async.py
@@ -211,6 +211,37 @@ async def main() -> None:
                     )
         # [END extract_invoice_fields]
 
+        # [START get_usage]
+        # Access usage details from the poller (available after result() completes).
+        # Usage reports resource consumption for billing estimation:
+        #
+        # - document_pages_standard/basic/minimal: Pages processed at each extraction tier.
+        #   Standard = layout + OCR (scanned docs), Basic = OCR only, Minimal = digital formats
+        #   (DOCX, XLSX, HTML, TXT) that need no OCR. Charged per 1,000 pages.
+        #
+        # - contextualization_tokens: Fixed-rate tokens charged by Content Understanding for
+        #   preparing context, generating confidence scores, source grounding, and formatting
+        #   output. Typically 1,000 tokens per page. Charged separately from LLM tokens.
+        #
+        # - tokens: Dict of "{model}-input" / "{model}-output" token counts consumed by your
+        #   Foundry model deployment (e.g. "gpt-4.1-input", "gpt-4.1-output"). These are
+        #   billed on your Foundry deployment, not on Content Understanding.
+        #
+        # For full pricing details, see:
+        # https://learn.microsoft.com/azure/ai-services/content-understanding/pricing-explainer
+        usage = poller.usage
+        if usage:
+            print("\nUsage Details:")
+            if usage.document_pages_standard is not None:
+                print(f"  Document pages (standard): {usage.document_pages_standard}")
+            if usage.contextualization_tokens is not None:
+                print(f"  Contextualization tokens: {usage.contextualization_tokens}")
+            if usage.tokens:
+                print("  Model tokens:")
+                for model, count in usage.tokens.items():
+                    print(f"    {model}: {count}")
+        # [END get_usage]
+
         if not isinstance(credential, AzureKeyCredential):
             await credential.close()
 
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_analyze_url_async.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_analyze_url_async.py
index 04562c733718..e4cca23f5474 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_analyze_url_async.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_analyze_url_async.py
@@ -74,7 +74,7 @@ async def main() -> None:
         # You can replace this URL with your own publicly accessible document URL.
         document_url = "https://raw.githubusercontent.com/Azure-Samples/azure-ai-content-understanding-assets/main/document/mixed_financial_docs.pdf"
 
-        print(f"Analyzing document from URL with prebuilt-documentSearch...")
+        print("Analyzing document from URL with prebuilt-documentSearch...")
         print(f"  URL: {document_url}")
 
         poller = await client.begin_analyze(
@@ -119,7 +119,7 @@ async def main() -> None:
 
         range_doc_content = cast(DocumentContent, range_result.contents[0])
         print(
-            f"Content range analysis returned pages"
+            "Content range analysis returned pages"
             f" {range_doc_content.start_page_number} - {range_doc_content.end_page_number}"
         )
 
@@ -133,7 +133,7 @@ async def main() -> None:
 
         combine_doc_content = cast(DocumentContent, combine_range_result.contents[0])
         print(
-            f"Combined content range analysis returned pages"
+            "Combined content range analysis returned pages"
             f" {combine_doc_content.start_page_number} - {combine_doc_content.end_page_number}"
         )
         # [END analyze_document_url_with_content_range]
@@ -144,7 +144,7 @@ async def main() -> None:
         print("=" * 60)
         video_url = "https://raw.githubusercontent.com/Azure-Samples/azure-ai-content-understanding-assets/main/videos/sdk_samples/FlightSimulator.mp4"
 
-        print(f"Analyzing video from URL with prebuilt-videoSearch...")
+        print("Analyzing video from URL with prebuilt-videoSearch...")
         print(f"  URL: {video_url}")
 
         poller = await client.begin_analyze(
@@ -197,7 +197,7 @@ async def main() -> None:
         for range_media in video_range_result.contents:
             range_video_content = cast(AudioVisualContent, range_media)
             print(
-                f"Content range segment:"
+                "Content range segment:"
                 f" {range_video_content.start_time_ms} ms - {range_video_content.end_time_ms} ms"
             )
         # [END analyze_video_url_with_content_range]
@@ -220,7 +220,7 @@ async def main() -> None:
         for from_media in video_from_result.contents:
             from_video = cast(AudioVisualContent, from_media)
             print(
-                f"'10000-' segment:"
+                "'10000-' segment:"
                 f" {from_video.start_time_ms} ms - {from_video.end_time_ms} ms"
             )
 
@@ -239,7 +239,7 @@ async def main() -> None:
         for subsec_media in video_subsec_result.contents:
             subsec_video = cast(AudioVisualContent, subsec_media)
             print(
-                f"'1200-3651' segment:"
+                "'1200-3651' segment:"
                 f" {subsec_video.start_time_ms} ms - {subsec_video.end_time_ms} ms"
             )
 
@@ -258,7 +258,7 @@ async def main() -> None:
         for combine_media in video_combine_result.contents:
             combine_video = cast(AudioVisualContent, combine_media)
             print(
-                f"'0-3000,30000-' segment:"
+                "'0-3000,30000-' segment:"
                 f" {combine_video.start_time_ms} ms - {combine_video.end_time_ms} ms"
             )
         # [END analyze_video_url_with_additional_content_ranges]
@@ -269,7 +269,7 @@ async def main() -> None:
         print("=" * 60)
         audio_url = "https://raw.githubusercontent.com/Azure-Samples/azure-ai-content-understanding-assets/main/audio/callCenterRecording.mp3"
 
-        print(f"Analyzing audio from URL with prebuilt-audioSearch...")
+        print("Analyzing audio from URL with prebuilt-audioSearch...")
         print(f"  URL: {audio_url}")
 
         poller = await client.begin_analyze(
@@ -316,7 +316,7 @@ async def main() -> None:
 
         range_audio_content = cast(AudioVisualContent, audio_range_result.contents[0])
         print(
-            f"Content range audio segment:"
+            "Content range audio segment:"
             f" {range_audio_content.start_time_ms} ms - {range_audio_content.end_time_ms} ms"
         )
         # [END analyze_audio_url_with_content_range]
@@ -338,7 +338,7 @@ async def main() -> None:
         audio_from_result = await audio_from_poller.result()
         audio_from_content = cast(AudioVisualContent, audio_from_result.contents[0])
         print(
-            f"'10000-':"
+            "'10000-':"
             f" {audio_from_content.start_time_ms} ms - {audio_from_content.end_time_ms} ms"
         )
 
@@ -356,7 +356,7 @@ async def main() -> None:
         audio_subsec_result = await audio_subsec_poller.result()
         audio_subsec_content = cast(AudioVisualContent, audio_subsec_result.contents[0])
         print(
-            f"'1200-3651':"
+            "'1200-3651':"
             f" {audio_subsec_content.start_time_ms} ms - {audio_subsec_content.end_time_ms} ms"
         )
 
@@ -374,7 +374,7 @@ async def main() -> None:
         audio_combine_result = await audio_combine_poller.result()
         audio_combine_content = cast(AudioVisualContent, audio_combine_result.contents[0])
         print(
-            f"'0-3000,30000-':"
+            "'0-3000,30000-':"
             f" {audio_combine_content.start_time_ms} ms - {audio_combine_content.end_time_ms} ms"
         )
         # [END analyze_audio_url_with_additional_content_ranges]
@@ -385,7 +385,7 @@ async def main() -> None:
         print("=" * 60)
         image_url = "https://raw.githubusercontent.com/Azure-Samples/azure-ai-content-understanding-assets/main/image/pieChart.jpg"
 
-        print(f"Analyzing image from URL with prebuilt-imageSearch...")
+        print("Analyzing image from URL with prebuilt-imageSearch...")
         print(f"  URL: {image_url}")
 
         poller = await client.begin_analyze(
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_copy_analyzer_async.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_copy_analyzer_async.py
index 3878d7e67eff..80a30a65b565 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_copy_analyzer_async.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_copy_analyzer_async.py
@@ -119,7 +119,7 @@ async def main() -> None:
         )
         await poller.result()
 
-        print(f"Analyzer copied successfully!")
+        print("Analyzer copied successfully!")
         # [END copy_analyzer]
 
         # [START update_and_verify_analyzer]
@@ -133,7 +133,7 @@ async def main() -> None:
             tags={"modelType": "model_in_production"},
         )
 
-        print(f"Updating target analyzer with production tag...")
+        print("Updating target analyzer with production tag...")
         await client.update_analyzer(
             analyzer_id=target_analyzer_id, resource=updated_analyzer
         )
@@ -148,7 +148,7 @@ async def main() -> None:
         # [END update_and_verify_analyzer]
 
         # [START delete_copied_analyzers]
-        print(f"\nCleaning up analyzers...")
+        print("\nCleaning up analyzers...")
 
         try:
             await client.delete_analyzer(analyzer_id=source_analyzer_id)
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_grant_copy_auth_async.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_grant_copy_auth_async.py
index 59e83da8c893..d9f2fa896490 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_grant_copy_auth_async.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/async_samples/sample_grant_copy_auth_async.py
@@ -75,8 +75,8 @@
     Example resource ID format:
     /subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.CognitiveServices/accounts/{name}
 
-    Important: Cross-resource copying requires credential-based authentication (such as DefaultAzureCredential).
-    API keys cannot be used for cross-resource operations.
+    Note: If API keys are not provided, DefaultAzureCredential will be used.
+    Cross-resource copying with DefaultAzureCredential requires 'Cognitive Services User' role on both source and target resources.
 """
 
 import asyncio
@@ -208,7 +208,7 @@ async def main() -> None:
                 resource=source_analyzer,
             )
             await poller.result()
-            print(f"  Source analyzer created successfully!")
+            print("  Source analyzer created successfully!")
 
             # Step 2: Grant copy authorization
             # Authorization must be granted by the source resource before the target resource can copy
@@ -216,7 +216,7 @@ async def main() -> None:
             # - The source analyzer ID to be copied
             # - The target Azure resource ID that is allowed to receive the copy
             # - The target region where the copy will be performed (optional, defaults to current region)
-            print(f"\nStep 2: Granting copy authorization from source resource...")
+            print("\nStep 2: Granting copy authorization from source resource...")
             print(f"  Target Azure Resource ID: {target_resource_id}")
             print(f"  Target Region: {target_region}")
 
@@ -226,7 +226,7 @@ async def main() -> None:
                 target_region=target_region,
             )
 
-            print(f"  Authorization granted successfully!")
+            print("  Authorization granted successfully!")
             print(f"  Target Azure Resource ID: {copy_auth.target_azure_resource_id}")
             print(f"  Target Region: {target_region}")
             print(f"  Expires at: {copy_auth.expires_at}")
@@ -235,7 +235,7 @@ async def main() -> None:
             # The copy_analyzer method must be called on the target client because the target
             # resource is the one receiving and creating the copy. The target resource validates
             # that authorization was previously granted by the source resource.
-            print(f"\nStep 3: Copying analyzer from source to target...")
+            print("\nStep 3: Copying analyzer from source to target...")
             print(f"  Source Analyzer ID: {source_analyzer_id}")
             print(f"  Source Azure Resource ID: {source_resource_id}")
             print(f"  Source Region: {source_region}")
@@ -248,22 +248,22 @@ async def main() -> None:
                 source_region=source_region,
             )
             await copy_poller.result()
-            print(f"  Analyzer copied successfully to target resource!")
+            print("  Analyzer copied successfully to target resource!")
 
             # Step 4: Verify the copy
             # Retrieve the analyzer from the target resource to verify the copy was successful
-            print(f"\nStep 4: Verifying the copied analyzer...")
+            print("\nStep 4: Verifying the copied analyzer...")
             copied_analyzer = await target_client.get_analyzer(
                 analyzer_id=target_analyzer_id
             )
             print(f"  Target Analyzer ID: {copied_analyzer.analyzer_id}")
             print(f"  Description: {copied_analyzer.description}")
             print(f"  Status: {copied_analyzer.status}")
-            print(f"\nCross-resource copy completed successfully!")
+            print("\nCross-resource copy completed successfully!")
 
     finally:
         # Clean up - create new client instances for cleanup since the original ones are closed
-        print(f"\nCleaning up...")
+        print("\nCleaning up...")
         cleanup_source_client = ContentUnderstandingClient(
             endpoint=source_endpoint, credential=source_credential
         )
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_analyze_invoice.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_analyze_invoice.py
index 38542a8a3fa8..c7812c180597 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_analyze_invoice.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_analyze_invoice.py
@@ -205,6 +205,37 @@ def main() -> None:
                 )
     # [END extract_invoice_fields]
 
+    # [START get_usage]
+    # Access usage details from the poller (available after result() completes).
+    # Usage reports resource consumption for billing estimation:
+    #
+    # - document_pages_standard/basic/minimal: Pages processed at each extraction tier.
+    #   Standard = layout + OCR (scanned docs), Basic = OCR only, Minimal = digital formats
+    #   (DOCX, XLSX, HTML, TXT) that need no OCR. Charged per 1,000 pages.
+    #
+    # - contextualization_tokens: Fixed-rate tokens charged by Content Understanding for
+    #   preparing context, generating confidence scores, source grounding, and formatting
+    #   output. Typically 1,000 tokens per page. Charged separately from LLM tokens.
+    #
+    # - tokens: Dict of "{model}-input" / "{model}-output" token counts consumed by your
+    #   Foundry model deployment (e.g. "gpt-4.1-input", "gpt-4.1-output"). These are
+    #   billed on your Foundry deployment, not on Content Understanding.
+    #
+    # For full pricing details, see:
+    # https://learn.microsoft.com/azure/ai-services/content-understanding/pricing-explainer
+    usage = poller.usage
+    if usage:
+        print("\nUsage Details:")
+        if usage.document_pages_standard is not None:
+            print(f"  Document pages (standard): {usage.document_pages_standard}")
+        if usage.contextualization_tokens is not None:
+            print(f"  Contextualization tokens: {usage.contextualization_tokens}")
+        if usage.tokens:
+            print("  Model tokens:")
+            for model, count in usage.tokens.items():
+                print(f"    {model}: {count}")
+    # [END get_usage]
+
 
 if __name__ == "__main__":
     main()
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_analyze_url.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_analyze_url.py
index 1be6e13c0073..0c3e9570d098 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_analyze_url.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_analyze_url.py
@@ -72,7 +72,7 @@ def main() -> None:
     # You can replace this URL with your own publicly accessible document URL.
     document_url = "https://raw.githubusercontent.com/Azure-Samples/azure-ai-content-understanding-assets/main/document/mixed_financial_docs.pdf"
 
-    print(f"Analyzing document from URL with prebuilt-documentSearch...")
+    print("Analyzing document from URL with prebuilt-documentSearch...")
     print(f"  URL: {document_url}")
 
     poller = client.begin_analyze(
@@ -117,7 +117,7 @@ def main() -> None:
 
     range_doc_content = cast(DocumentContent, range_result.contents[0])
     print(
-        f"Content range analysis returned pages"
+        "Content range analysis returned pages"
         f" {range_doc_content.start_page_number} - {range_doc_content.end_page_number}"
     )
 
@@ -131,7 +131,7 @@ def main() -> None:
 
     combine_doc_content = cast(DocumentContent, combine_range_result.contents[0])
     print(
-        f"Combined content range analysis returned pages"
+        "Combined content range analysis returned pages"
         f" {combine_doc_content.start_page_number} - {combine_doc_content.end_page_number}"
     )
     # [END analyze_document_url_with_content_range]
@@ -142,7 +142,7 @@ def main() -> None:
     print("=" * 60)
     video_url = "https://raw.githubusercontent.com/Azure-Samples/azure-ai-content-understanding-assets/main/videos/sdk_samples/FlightSimulator.mp4"
 
-    print(f"Analyzing video from URL with prebuilt-videoSearch...")
+    print("Analyzing video from URL with prebuilt-videoSearch...")
     print(f"  URL: {video_url}")
 
     poller = client.begin_analyze(
@@ -193,7 +193,7 @@ def main() -> None:
     for range_media in video_range_result.contents:
         range_video_content = cast(AudioVisualContent, range_media)
         print(
-            f"Content range segment:"
+            "Content range segment:"
             f" {range_video_content.start_time_ms} ms - {range_video_content.end_time_ms} ms"
         )
     # [END analyze_video_url_with_content_range]
@@ -216,7 +216,7 @@ def main() -> None:
     for from_media in video_from_result.contents:
         from_video = cast(AudioVisualContent, from_media)
         print(
-            f"'10000-' segment:"
+            "'10000-' segment:"
             f" {from_video.start_time_ms} ms - {from_video.end_time_ms} ms"
         )
 
@@ -235,7 +235,7 @@ def main() -> None:
     for subsec_media in video_subsec_result.contents:
         subsec_video = cast(AudioVisualContent, subsec_media)
         print(
-            f"'1200-3651' segment:"
+            "'1200-3651' segment:"
             f" {subsec_video.start_time_ms} ms - {subsec_video.end_time_ms} ms"
         )
 
@@ -254,7 +254,7 @@ def main() -> None:
     for combine_media in video_combine_result.contents:
         combine_video = cast(AudioVisualContent, combine_media)
         print(
-            f"'0-3000,30000-' segment:"
+            "'0-3000,30000-' segment:"
             f" {combine_video.start_time_ms} ms - {combine_video.end_time_ms} ms"
         )
     # [END analyze_video_url_with_additional_content_ranges]
@@ -265,7 +265,7 @@ def main() -> None:
     print("=" * 60)
     audio_url = "https://raw.githubusercontent.com/Azure-Samples/azure-ai-content-understanding-assets/main/audio/callCenterRecording.mp3"
 
-    print(f"Analyzing audio from URL with prebuilt-audioSearch...")
+    print("Analyzing audio from URL with prebuilt-audioSearch...")
     print(f"  URL: {audio_url}")
 
     poller = client.begin_analyze(
@@ -309,7 +309,7 @@ def main() -> None:
 
     range_audio_content = cast(AudioVisualContent, audio_range_result.contents[0])
     print(
-        f"Content range audio segment:"
+        "Content range audio segment:"
         f" {range_audio_content.start_time_ms} ms - {range_audio_content.end_time_ms} ms"
     )
     # [END analyze_audio_url_with_content_range]
@@ -331,7 +331,7 @@ def main() -> None:
     audio_from_result = audio_from_poller.result()
     audio_from_content = cast(AudioVisualContent, audio_from_result.contents[0])
     print(
-        f"'10000-':"
+        "'10000-':"
         f" {audio_from_content.start_time_ms} ms - {audio_from_content.end_time_ms} ms"
     )
 
@@ -349,7 +349,7 @@ def main() -> None:
     audio_subsec_result = audio_subsec_poller.result()
     audio_subsec_content = cast(AudioVisualContent, audio_subsec_result.contents[0])
     print(
-        f"'1200-3651':"
+        "'1200-3651':"
         f" {audio_subsec_content.start_time_ms} ms - {audio_subsec_content.end_time_ms} ms"
     )
 
@@ -367,7 +367,7 @@ def main() -> None:
     audio_combine_result = audio_combine_poller.result()
     audio_combine_content = cast(AudioVisualContent, audio_combine_result.contents[0])
     print(
-        f"'0-3000,30000-':"
+        "'0-3000,30000-':"
         f" {audio_combine_content.start_time_ms} ms - {audio_combine_content.end_time_ms} ms"
     )
     # [END analyze_audio_url_with_additional_content_ranges]
@@ -378,7 +378,7 @@ def main() -> None:
     print("=" * 60)
     image_url = "https://raw.githubusercontent.com/Azure-Samples/azure-ai-content-understanding-assets/main/image/pieChart.jpg"
 
-    print(f"Analyzing image from URL with prebuilt-imageSearch...")
+    print("Analyzing image from URL with prebuilt-imageSearch...")
     print(f"  URL: {image_url}")
 
     poller = client.begin_analyze(
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_copy_analyzer.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_copy_analyzer.py
index ceae81e28485..c023cb6c874f 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_copy_analyzer.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_copy_analyzer.py
@@ -116,7 +116,7 @@ def main() -> None:
     )
     poller.result()
 
-    print(f"Analyzer copied successfully!")
+    print("Analyzer copied successfully!")
     # [END copy_analyzer]
 
     # [START update_and_verify_analyzer]
@@ -130,7 +130,7 @@ def main() -> None:
         tags={"modelType": "model_in_production"},
     )
 
-    print(f"Updating target analyzer with production tag...")
+    print("Updating target analyzer with production tag...")
     client.update_analyzer(analyzer_id=target_analyzer_id, resource=updated_analyzer)
 
     # Verify the update
@@ -143,7 +143,7 @@ def main() -> None:
     # [END update_and_verify_analyzer]
 
     # [START delete_copied_analyzers]
-    print(f"\nCleaning up analyzers...")
+    print("\nCleaning up analyzers...")
 
     try:
         client.delete_analyzer(analyzer_id=source_analyzer_id)
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_grant_copy_auth.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_grant_copy_auth.py
index 17cd03af239c..b4ef028c34c4 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_grant_copy_auth.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/samples/sample_grant_copy_auth.py
@@ -75,8 +75,8 @@
     Example resource ID format:
     /subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.CognitiveServices/accounts/{name}
 
-    Important: Cross-resource copying requires credential-based authentication (such as DefaultAzureCredential).
-    API keys cannot be used for cross-resource operations.
+    Note: If API keys are not provided, DefaultAzureCredential will be used.
+    Cross-resource copying with DefaultAzureCredential requires 'Cognitive Services User' role on both source and target resources.
 """
 
 import os
@@ -206,7 +206,7 @@ def main() -> None:
             resource=source_analyzer,
         )
         poller.result()
-        print(f"  Source analyzer created successfully!")
+        print("  Source analyzer created successfully!")
 
         # Step 2: Grant copy authorization
         # Authorization must be granted by the source resource before the target resource can copy
@@ -214,7 +214,7 @@ def main() -> None:
         # - The source analyzer ID to be copied
         # - The target Azure resource ID that is allowed to receive the copy
         # - The target region where the copy will be performed (optional, defaults to current region)
-        print(f"\nStep 2: Granting copy authorization from source resource...")
+        print("\nStep 2: Granting copy authorization from source resource...")
         print(f"  Target Azure Resource ID: {target_resource_id}")
         print(f"  Target Region: {target_region}")
 
@@ -224,7 +224,7 @@ def main() -> None:
             target_region=target_region,
         )
 
-        print(f"  Authorization granted successfully!")
+        print("  Authorization granted successfully!")
         print(f"  Target Azure Resource ID: {copy_auth.target_azure_resource_id}")
         print(f"  Target Region: {target_region}")
         print(f"  Expires at: {copy_auth.expires_at}")
@@ -233,7 +233,7 @@ def main() -> None:
         # The copy_analyzer method must be called on the target client because the target
         # resource is the one receiving and creating the copy. The target resource validates
         # that authorization was previously granted by the source resource.
-        print(f"\nStep 3: Copying analyzer from source to target...")
+        print("\nStep 3: Copying analyzer from source to target...")
         print(f"  Source Analyzer ID: {source_analyzer_id}")
         print(f"  Source Azure Resource ID: {source_resource_id}")
         print(f"  Source Region: {source_region}")
@@ -246,20 +246,20 @@ def main() -> None:
             source_region=source_region,
         )
         copy_poller.result()
-        print(f"  Analyzer copied successfully to target resource!")
+        print("  Analyzer copied successfully to target resource!")
 
         # Step 4: Verify the copy
         # Retrieve the analyzer from the target resource to verify the copy was successful
-        print(f"\nStep 4: Verifying the copied analyzer...")
+        print("\nStep 4: Verifying the copied analyzer...")
         copied_analyzer = target_client.get_analyzer(analyzer_id=target_analyzer_id)
         print(f"  Target Analyzer ID: {copied_analyzer.analyzer_id}")
         print(f"  Description: {copied_analyzer.description}")
         print(f"  Status: {copied_analyzer.status}")
-        print(f"\nCross-resource copy completed successfully!")
+        print("\nCross-resource copy completed successfully!")
 
     finally:
         # Clean up: Delete both source and target analyzers
-        print(f"\nCleaning up...")
+        print("\nCleaning up...")
         try:
             source_client.delete_analyzer(analyzer_id=source_analyzer_id)
             print(f"  Source analyzer '{source_analyzer_id}' deleted.")
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/samples/test_sample_analyze_invoice.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/samples/test_sample_analyze_invoice.py
index 865815bd13cf..97015d1e95cc 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/samples/test_sample_analyze_invoice.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/samples/test_sample_analyze_invoice.py
@@ -215,4 +215,26 @@ def test_sample_analyze_invoice(self, contentunderstanding_endpoint: str, **kwar
         else:
             print("[INFO] LineItems field not found in this document")
 
+        # Verify usage details (available after result() completes)
+        usage = poller.usage
+        assert usage is not None, "Usage details should be available after analysis"
+        print("[PASS] Usage details available")
+
+        # Verify at least one page metric is set
+        has_pages = (
+            (usage.document_pages_standard is not None and usage.document_pages_standard > 0)
+            or (usage.document_pages_basic is not None and usage.document_pages_basic > 0)
+            or (usage.document_pages_minimal is not None and usage.document_pages_minimal > 0)
+        )
+        assert has_pages, "Usage should report at least one document page metric"
+        print(f"[PASS] Usage reports document pages (standard={usage.document_pages_standard}, basic={usage.document_pages_basic}, minimal={usage.document_pages_minimal})")
+
+        # Verify token usage is reported
+        if usage.tokens:
+            assert len(usage.tokens) > 0, "Token usage should have at least one model entry"
+            for model, count in usage.tokens.items():
+                assert count > 0, f"Token count for {model} should be positive"
+                print(f"[INFO] Token usage: {model} = {count}")
+            print(f"[PASS] Token usage reported for {len(usage.tokens)} model(s)")
+
         print("\n[SUCCESS] All test_sample_analyze_invoice assertions passed")
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/samples/test_sample_analyze_invoice_async.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/samples/test_sample_analyze_invoice_async.py
index 1015b54454a9..d041c091064a 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/samples/test_sample_analyze_invoice_async.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/samples/test_sample_analyze_invoice_async.py
@@ -214,5 +214,27 @@ async def test_sample_analyze_invoice_async(self, contentunderstanding_endpoint:
         else:
             print("[INFO] LineItems field not found in this document")
 
+        # Verify usage details (available after result() completes)
+        usage = poller.usage
+        assert usage is not None, "Usage details should be available after analysis"
+        print("[PASS] Usage details available")
+
+        # Verify at least one page metric is set
+        has_pages = (
+            (usage.document_pages_standard is not None and usage.document_pages_standard > 0)
+            or (usage.document_pages_basic is not None and usage.document_pages_basic > 0)
+            or (usage.document_pages_minimal is not None and usage.document_pages_minimal > 0)
+        )
+        assert has_pages, "Usage should report at least one document page metric"
+        print(f"[PASS] Usage reports document pages (standard={usage.document_pages_standard}, basic={usage.document_pages_basic}, minimal={usage.document_pages_minimal})")
+
+        # Verify token usage is reported
+        if usage.tokens:
+            assert len(usage.tokens) > 0, "Token usage should have at least one model entry"
+            for model, count in usage.tokens.items():
+                assert count > 0, f"Token count for {model} should be positive"
+                print(f"[INFO] Token usage: {model} = {count}")
+            print(f"[PASS] Token usage reported for {len(usage.tokens)} model(s)")
+
         await client.close()
         print("\n[SUCCESS] All test_sample_analyze_invoice_async assertions passed")
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_analyzer_operation_id.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_analyzer_operation_id.py
index d3647016a6ae..c1e0a377b9d1 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_analyzer_operation_id.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_analyzer_operation_id.py
@@ -16,7 +16,7 @@
     AnalyzeLROPoller,
     _parse_operation_id,
 )
-from azure.ai.contentunderstanding.models import AnalysisInput
+from azure.ai.contentunderstanding.models import AnalysisInput, UsageDetails
 from azure.ai.contentunderstanding import ContentUnderstandingClient
 
 
@@ -164,3 +164,115 @@ def test_analyze_operation_returns_custom_poller(self):
         assert hasattr(result, "operation_id")
         operation_id = result.operation_id
         assert operation_id == "test-op-id-123"
+
+
+class TestAnalyzeLROPollerUsage:
+    """Test the usage property on AnalyzeLROPoller (GitHub issue #46249)."""
+
+    def _make_poller_with_usage(self, usage_json):
+        """Helper to create an AnalyzeLROPoller with a mocked final response containing usage data."""
+        mock_polling_method = Mock()
+        mock_initial_response = Mock()
+        mock_http_response = Mock()
+        mock_http_response.headers = {
+            "Operation-Location": "https://endpoint/contentunderstanding/analyzerResults/test-op-id?api-version=2025-11-01"
+        }
+        mock_initial_response.http_response = mock_http_response
+        mock_polling_method.return_value = mock_polling_method
+        mock_polling_method._initial_response = mock_initial_response
+
+        # Mock the final pipeline response (used by .usage property)
+        mock_final_response = Mock()
+        mock_final_http_response = Mock()
+        response_json = {
+            "id": "test-op-id",
+            "status": "Succeeded",
+            "result": {"contents": []},
+        }
+        if usage_json is not None:
+            response_json["usage"] = usage_json
+        mock_final_http_response.json.return_value = response_json
+        mock_final_response.http_response = mock_final_http_response
+        mock_polling_method._pipeline_response = mock_final_response
+
+        return AnalyzeLROPoller(
+            client=Mock(),
+            initial_response=Mock(),
+            deserialization_callback=Mock(),
+            polling_method=mock_polling_method,
+        )
+
+    def test_usage_with_full_data(self):
+        """Test usage property returns UsageDetails with all fields populated."""
+        usage_json = {
+            "documentPagesMinimal": 0,
+            "documentPagesBasic": 0,
+            "documentPagesStandard": 2,
+            "audioHours": 0.0,
+            "videoHours": 0.0,
+            "contextualizationTokens": 1500,
+            "tokens": {
+                "gpt-4.1-input": 500,
+                "gpt-4.1-output": 200,
+                "text-embedding-3-large-input": 300,
+            },
+        }
+
+        poller = self._make_poller_with_usage(usage_json)
+        usage = poller.usage
+
+        assert usage is not None
+        assert isinstance(usage, UsageDetails)
+        assert usage.document_pages_standard == 2
+        assert usage.document_pages_minimal == 0
+        assert usage.document_pages_basic == 0
+        assert usage.contextualization_tokens == 1500
+        assert usage.tokens == {
+            "gpt-4.1-input": 500,
+            "gpt-4.1-output": 200,
+            "text-embedding-3-large-input": 300,
+        }
+
+    def test_usage_with_partial_data(self):
+        """Test usage property with only some fields populated."""
+        usage_json = {
+            "documentPagesStandard": 1,
+            "contextualizationTokens": 100,
+        }
+
+        poller = self._make_poller_with_usage(usage_json)
+        usage = poller.usage
+
+        assert usage is not None
+        assert usage.document_pages_standard == 1
+        assert usage.contextualization_tokens == 100
+        assert usage.document_pages_minimal is None
+        assert usage.tokens is None
+
+    def test_usage_returns_none_when_not_present(self):
+        """Test usage property returns None when usage is not in the response."""
+        poller = self._make_poller_with_usage(None)
+        usage = poller.usage
+
+        assert usage is None
+
+    def test_usage_returns_none_before_completion(self):
+        """Test usage property returns None when the polling has not yet completed."""
+        mock_polling_method = Mock()
+
+        poller = AnalyzeLROPoller(
+            client=Mock(),
+            initial_response=Mock(),
+            deserialization_callback=Mock(),
+            polling_method=mock_polling_method,
+        )
+
+        # Simulate an in-progress operation: LROPoller.done() returns True when
+        # _thread is None, so set _thread to a mock with is_alive() returning True.
+        mock_thread = Mock()
+        mock_thread.is_alive.return_value = True
+        poller._thread = mock_thread
+
+        assert not poller.done()
+        usage = poller.usage
+        assert usage is None
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_analyzer_operation_id_async.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_analyzer_operation_id_async.py
new file mode 100644
index 000000000000..7766b1c4c0cc
--- /dev/null
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_analyzer_operation_id_async.py
@@ -0,0 +1,90 @@
+# pylint: disable=line-too-long,useless-suppression
+# coding=utf-8
+# --------------------------------------------------------------------------
+# Copyright (c) Microsoft Corporation. All rights reserved.
+# Licensed under the MIT License. See License.txt in the project root for license information.
+# --------------------------------------------------------------------------
+
+"""
+Tests for Content Understanding async analyzer operation ID and usage functionality.
+"""
+
+from unittest.mock import Mock
+from azure.ai.contentunderstanding.aio.models._patch import AnalyzeAsyncLROPoller
+from azure.ai.contentunderstanding.models import UsageDetails
+
+
+class TestAnalyzeAsyncLROPollerUsage:
+    """Test the usage property on AnalyzeAsyncLROPoller (async counterpart)."""
+
+    def _make_async_poller_with_usage(self, usage_json, done=True):
+        """Helper to create an AnalyzeAsyncLROPoller with a mocked final response."""
+        mock_polling_method = Mock()
+        mock_initial_response = Mock()
+        mock_http_response = Mock()
+        mock_http_response.headers = {
+            "Operation-Location": "https://endpoint/contentunderstanding/analyzerResults/test-op-id?api-version=2025-11-01"
+        }
+        mock_initial_response.http_response = mock_http_response
+        mock_polling_method.return_value = mock_polling_method
+        mock_polling_method._initial_response = mock_initial_response
+
+        mock_final_response = Mock()
+        mock_final_http_response = Mock()
+        response_json = {
+            "id": "test-op-id",
+            "status": "Succeeded",
+            "result": {"contents": []},
+        }
+        if usage_json is not None:
+            response_json["usage"] = usage_json
+        mock_final_http_response.json.return_value = response_json
+        mock_final_response.http_response = mock_final_http_response
+        mock_polling_method._pipeline_response = mock_final_response
+
+        poller = AnalyzeAsyncLROPoller(
+            client=Mock(),
+            initial_response=Mock(),
+            deserialization_callback=Mock(),
+            polling_method=mock_polling_method,
+        )
+        # AsyncLROPoller.done() checks self._done (defaults to False after __init__)
+        poller._done = done
+        return poller
+
+    def test_async_usage_with_full_data(self):
+        """Test async usage property returns UsageDetails with all fields populated."""
+        usage_json = {
+            "documentPagesMinimal": 0,
+            "documentPagesBasic": 0,
+            "documentPagesStandard": 2,
+            "audioHours": 0.0,
+            "videoHours": 0.0,
+            "contextualizationTokens": 1500,
+            "tokens": {
+                "gpt-4.1-input": 500,
+                "gpt-4.1-output": 200,
+            },
+        }
+
+        poller = self._make_async_poller_with_usage(usage_json)
+        usage = poller.usage
+
+        assert usage is not None
+        assert isinstance(usage, UsageDetails)
+        assert usage.document_pages_standard == 2
+        assert usage.contextualization_tokens == 1500
+        assert usage.tokens == {"gpt-4.1-input": 500, "gpt-4.1-output": 200}
+
+    def test_async_usage_returns_none_when_not_present(self):
+        """Test async usage property returns None when usage is not in the response."""
+        poller = self._make_async_poller_with_usage(None)
+        usage = poller.usage
+        assert usage is None
+
+    def test_async_usage_returns_none_before_completion(self):
+        """Test async usage property returns None when the polling has not yet completed."""
+        poller = self._make_async_poller_with_usage({"documentPagesStandard": 1}, done=False)
+        assert not poller.done()
+        usage = poller.usage
+        assert usage is None
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_content_understanding_content_analyzers_operations.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_content_understanding_content_analyzers_operations.py
index a7592b044843..242843293ab0 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_content_understanding_content_analyzers_operations.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_content_understanding_content_analyzers_operations.py
@@ -21,6 +21,7 @@
     new_simple_content_analyzer_object,
     new_marketing_video_analyzer_object,
     assert_poller_properties,
+    assert_analyze_poller_usage,
     assert_simple_content_analyzer_result,
     save_analysis_result_to_file,
     save_keyframe_image_to_file,
@@ -604,6 +605,7 @@ def test_content_analyzers_analyze_binary_extract_markdown(self, contentundersta
         # Wait for completion
         result = poller.result()
         assert_poller_properties(poller)
+        assert_analyze_poller_usage(poller)
 
         # Verify result
         assert result is not None, "Analysis result should not be null"
@@ -739,6 +741,7 @@ def test_content_analyzers_analyze_configs(self, contentunderstanding_endpoint:
         # Wait for completion
         result = poller.result()
         assert_poller_properties(poller)
+        assert_analyze_poller_usage(poller)
 
         # Verify result
         assert result is not None, "Analysis result should not be null"
@@ -801,6 +804,7 @@ def test_content_analyzers_analyze_return_raw_json(self, contentunderstanding_en
         # Wait for completion
         result = poller.result()
         assert_poller_properties(poller)
+        assert_analyze_poller_usage(poller)
 
         # Verify operation completed successfully
         assert result is not None, "Analysis result should not be null"
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_content_understanding_content_analyzers_operations_async.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_content_understanding_content_analyzers_operations_async.py
index e2f42ae04a9c..d5b8450ed9ab 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_content_understanding_content_analyzers_operations_async.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_content_understanding_content_analyzers_operations_async.py
@@ -21,6 +21,7 @@
     new_invoice_analyzer_object,
     new_marketing_video_analyzer_object,
     assert_poller_properties,
+    assert_analyze_poller_usage,
     assert_simple_content_analyzer_result,
     assert_invoice_fields,
     assert_document_properties,
@@ -978,6 +979,7 @@ async def test_analyze_binary_extract_markdown_async(self, contentunderstanding_
         # Wait for completion
         result = await poller.result()
         assert_poller_properties(poller)
+        assert_analyze_poller_usage(poller)
 
         # Verify result
         assert result is not None, "Analysis result should not be null"
@@ -1113,6 +1115,7 @@ async def test_analyze_configs_async(self, contentunderstanding_endpoint: str) -
         # Wait for completion
         result = await poller.result()
         assert_poller_properties(poller)
+        assert_analyze_poller_usage(poller)
 
         # Verify result
         assert result is not None, "Analysis result should not be null"
@@ -1175,6 +1178,7 @@ async def test_analyze_return_raw_json_async(self, contentunderstanding_endpoint
         # Wait for completion
         result = await poller.result()
         assert_poller_properties(poller)
+        assert_analyze_poller_usage(poller)
 
         # Verify operation completed successfully
         assert result is not None, "Analysis result should not be null"
diff --git a/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_helpers.py b/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_helpers.py
index 5d405339cabf..57e1d071c2c4 100644
--- a/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_helpers.py
+++ b/sdk/contentunderstanding/azure-ai-contentunderstanding/tests/test_helpers.py
@@ -132,6 +132,25 @@ def assert_poller_properties(poller: Any, poller_name: str = "Poller") -> None:
     print(f"{poller_name} properties verified successfully")
 
 
+def assert_analyze_poller_usage(poller: Any, poller_name: str = "Poller") -> None:
+    """Assert the usage property on a completed analyze poller (AnalyzeLROPoller / AnalyzeAsyncLROPoller).
+
+    Must be called AFTER poller.result() has been called, since usage is only available
+    once the LRO has completed and the final polling response contains the usage data.
+
+    Args:
+        poller: The AnalyzeLROPoller or AnalyzeAsyncLROPoller instance to validate
+        poller_name: Optional name for the poller in log messages
+
+    Raises:
+        AssertionError: If the usage property is not available or invalid
+    """
+    assert hasattr(poller, "usage"), f"{poller_name} should have a 'usage' property"
+    usage = poller.usage
+    assert usage is not None, f"{poller_name} usage should not be None after completion"
+    print(f"{poller_name} usage verified: {usage}")
+
+
 def assert_simple_content_analyzer_result(analysis_result: Any, result_name: str = "Analysis result") -> None:
     """Assert simple content analyzer result properties and field extraction.