fix: merging dev changes to main (#2068)

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: Roopan-Microsoft <168007406+Roopan-Microsoft@users.noreply.github.com>
Co-authored-by: Ajit Padhi <v-padhiajit@microsoft.com>
Co-authored-by: Roopan P M <v-roopanpm@microsoft.com>
Co-authored-by: Pavan-Microsoft <v-kupavan@microsoft.com>
Co-authored-by: Ross Smith <ross-p-smith@users.noreply.github.com>
Co-authored-by: gpickett <122489228+gpickett@users.noreply.github.com>
Co-authored-by: Francia Riesco <friesco@microsoft.com>
Co-authored-by: Francia Riesco <Fr4nc3@users.noreply.github.com>
Co-authored-by: Prajwal D C <v-dcprajwal@microsoft.com>
Co-authored-by: Harmanpreet-Microsoft <v-harmanprka@microsoft.com>
Co-authored-by: UtkarshMishra-Microsoft <v-utkamishra@microsoft.com>
Co-authored-by: Prasanjeet-Microsoft <v-singhprasa@microsoft.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Kiran-Siluveru-Microsoft <v-ksiluveru@microsoft.com>
Co-authored-by: Prashant-Microsoft <v-pmalusare@microsoft.com>
Co-authored-by: Rohini-Microsoft <v-rwalunj@microsoft.com>
Co-authored-by: Avijit-Microsoft <v-aghorui@microsoft.com>
Co-authored-by: RaviKiran-Microsoft <v-ravikirans@microsoft.com>
Co-authored-by: Somesh Joshi <v-somejoshi@microsoft.com>
Co-authored-by: Himanshi Agrawal <v-himagrawal@microsoft.com>
Co-authored-by: pradeepjha-microsoft <v-pradeepjha@microsoft.com>
Co-authored-by: Harmanpreet Kaur <v-harmanpkau@microsoft.com>
Co-authored-by: Bangarraju-Microsoft <v-golib@microsoft.com>
Co-authored-by: Harsh-Microsoft <v-hbangera@microsoft.com>
Co-authored-by: Kanchan-Microsoft <v-knagshetti@microsoft.com>
Co-authored-by: Cristopher Coronado <cristofima@hotmail.com>
Co-authored-by: Cristopher Coronado Moreira <crcorona@pichincha.com>
Co-authored-by: Vamshi-Microsoft <v-vamolla@microsoft.com>
Co-authored-by: Thanusree-Microsoft <168087422+Thanusree-Microsoft@users.noreply.github.com>
Co-authored-by: Niraj Chaudhari (Persistent Systems Inc) <v-nirajcha@microsoft.com>
Co-authored-by: Rohini-Microsoft <168007985+Rohini-Microsoft@users.noreply.github.com>
Co-authored-by: Kingshuk-Microsoft <v-kidatta@microsoft.com>
Co-authored-by: Ayaz-Microsoft <v-ayazkhan@microsoft.com>
Co-authored-by: Abdul-Microsoft <v-amujeebta@microsoft.com>
Co-authored-by: Prekshith-Microsoft <v-pdj@microsoft.com>
Co-authored-by: Rafi-Microsoft <v-rafmd@microsoft.com>

Author: 34 people

.github/workflows/broken-links-checker.yml

@@ -35,7 +35,7 @@ jobs:
         uses: lycheeverse/lychee-action@v2.7.0
         with:
           args: >
-            --verbose --exclude-mail --no-progress --exclude ^https?://
+            --verbose --no-progress --exclude ^https?://
             ${{ steps.changed-markdown-files.outputs.all_changed_files }}
           failIfEmpty: false
         env:
@@ -48,7 +48,7 @@ jobs:
         uses: lycheeverse/lychee-action@v2.7.0
         with:
           args: >
-            --verbose --exclude-mail --no-progress --exclude ^https?://
+            --verbose --no-progress --exclude ^https?://
             '**/*.md'
           failIfEmpty: false
         env:

azure.yaml

@@ -56,7 +56,7 @@ services:
       prepackage:
         windows:
           shell: pwsh
-          run: poetry export -o requirements.txt; pip install -r requirements.txt
+          run: poetry export -o requirements.txt --without-hashes; pip install -r requirements.txt
         posix:
           shell: sh
-          run: poetry export -o requirements.txt; pip install -r requirements.txt
+          run: poetry export -o requirements.txt --without-hashes; pip install -r requirements.txt

code/backend/batch/host.json

@@ -10,6 +10,6 @@
   },
   "extensionBundle": {
     "id": "Microsoft.Azure.Functions.ExtensionBundle",
-    "version": "[3.*, 4.0.0)"
+    "version": "[4.*, 5.0.0)"
   }
-}
+}

code/backend/batch/utilities/helpers/llm_helper.py

@@ -91,26 +91,44 @@ def get_streaming_llm(self):
             )
 
     def get_embedding_model(self):
+        # Only pass dimensions for models that support it (text-embedding-3-*)
+        # text-embedding-ada-002 does NOT support the dimensions parameter
+        supports_dimensions = "text-embedding-3" in self.embedding_model.lower()
+        dimensions = None
+        if supports_dimensions and self.env_helper.AZURE_SEARCH_DIMENSIONS:
+            dimensions = int(self.env_helper.AZURE_SEARCH_DIMENSIONS)
+
         if self.auth_type_keys:
             return AzureOpenAIEmbeddings(
                 azure_endpoint=self.env_helper.AZURE_OPENAI_ENDPOINT,
                 api_key=self.env_helper.OPENAI_API_KEY,
                 azure_deployment=self.embedding_model,
+                dimensions=dimensions,
                 chunk_size=1,
             )
         else:
             return AzureOpenAIEmbeddings(
                 azure_endpoint=self.env_helper.AZURE_OPENAI_ENDPOINT,
                 azure_deployment=self.embedding_model,
+                dimensions=dimensions,
                 chunk_size=1,
                 azure_ad_token_provider=self.token_provider,
             )
 
     def generate_embeddings(self, input: Union[str, list[int]]) -> List[float]:
+        # Only pass dimensions for models that support it (text-embedding-3-*)
+        # text-embedding-ada-002 does NOT support the dimensions parameter
+        supports_dimensions = "text-embedding-3" in self.embedding_model.lower()
+        dimensions = None
+        if supports_dimensions and self.env_helper.AZURE_SEARCH_DIMENSIONS:
+            dimensions = int(self.env_helper.AZURE_SEARCH_DIMENSIONS)
+
+        kwargs = {"input": [input], "model": self.embedding_model}
+        if dimensions is not None:
+            kwargs["dimensions"] = dimensions
+
         return (
-            self.openai_client.embeddings.create(
-                input=[input], model=self.embedding_model
-            )
+            self.openai_client.embeddings.create(**kwargs)
             .data[0]
             .embedding
         )

code/tests/functional/app_config.py

@@ -38,7 +38,7 @@ class AppConfig:
         "AZURE_SEARCH_CONVERSATIONS_LOG_INDEX": "some-log-index",
         "AZURE_SEARCH_CONTENT_COLUMN": "content",
         "AZURE_SEARCH_CONTENT_VECTOR_COLUMN": "some-search-content-vector-columns",
-        "AZURE_SEARCH_DIMENSIONS": "some-search-dimensions",
+        "AZURE_SEARCH_DIMENSIONS": "1536",
         "AZURE_SEARCH_ENABLE_IN_DOMAIN": "True",
         "AZURE_SEARCH_FIELDS_ID": "some-search-fields-id",
         "AZURE_SEARCH_FIELDS_METADATA": "some-search-fields-metadata",

code/tests/utilities/helpers/test_llm_helper.py

@@ -20,6 +20,9 @@
 PROMPT_FLOW_DEPLOYMENT_NAME = "mock-deployment-name"
 
 
+AZURE_SEARCH_DIMENSIONS = "1536"
+
+
 @pytest.fixture(autouse=True)
 def env_helper_mock():
     with patch("backend.batch.utilities.helpers.llm_helper.EnvHelper") as mock:
@@ -36,6 +39,7 @@ def env_helper_mock():
         env_helper.AZURE_ML_WORKSPACE_NAME = AZURE_ML_WORKSPACE_NAME
         env_helper.PROMPT_FLOW_ENDPOINT_NAME = PROMPT_FLOW_ENDPOINT_NAME
         env_helper.PROMPT_FLOW_DEPLOYMENT_NAME = PROMPT_FLOW_DEPLOYMENT_NAME
+        env_helper.AZURE_SEARCH_DIMENSIONS = AZURE_SEARCH_DIMENSIONS
 
         yield env_helper
 

docs/customizing_azd_parameters.md

@@ -42,6 +42,7 @@ By default this template will use the environment name as the prefix to prevent
 | `AZURE_OPENAI_EMBEDDING_MODEL_NAME` | string | `text-embedding-ada-002` | Actual embedding model name |
 | `AZURE_OPENAI_EMBEDDING_MODEL_VERSION` | string | `2` | Embedding model version |
 | `AZURE_OPENAI_EMBEDDING_MODEL_CAPACITY` | integer | `100` | Embedding model capacity (TPM in thousands) |
+| `AZURE_SEARCH_DIMENSIONS` | integer | `1536` | Azure Search vector dimensions(Update dimensions for CosmosDB) |
 | `USE_ADVANCED_IMAGE_PROCESSING` | boolean | `false` | Enable vision LLM and Computer Vision for images (must be false for PostgreSQL) |
 | `ADVANCED_IMAGE_PROCESSING_MAX_IMAGES` | integer | `1` | Maximum images per vision model request |
 | `AZURE_OPENAI_VISION_MODEL` | string | `gpt-4.1` | Vision model deployment name |
@@ -90,5 +91,6 @@ azd env set AZURE_OPENAI_EMBEDDING_MODEL_CAPACITY 150
    - `AZURE_SEARCH_USE_INTEGRATED_VECTORIZATION=false`
    - `USE_ADVANCED_IMAGE_PROCESSING=false`
    - `ORCHESTRATION_STRATEGY=semantic_kernel` (recommended)
+   - `AZURE_SEARCH_DIMENSIONS=1536` (recommended)
 
 2. **Region Compatibility**: Not all services are available in all regions. Verify service availability in your chosen region before deployment.

infra/main.bicep

@@ -150,18 +150,6 @@ param useAdvancedImageProcessing bool = false
 @description('Optional. The maximum number of images to pass to the vision model in a single request.')
 param advancedImageProcessingMaxImages int = 1
 
-@description('Optional. Azure OpenAI Vision Model Deployment Name.')
-param azureOpenAIVisionModel string = 'gpt-4.1'
-
-@description('Optional. Azure OpenAI Vision Model Name.')
-param azureOpenAIVisionModelName string = 'gpt-4.1'
-
-@description('Optional. Azure OpenAI Vision Model Version.')
-param azureOpenAIVisionModelVersion string = '2025-04-14'
-
-@description('Optional. Azure OpenAI Vision Model Capacity - See here for more info  https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/quota.')
-param azureOpenAIVisionModelCapacity int = 10
-
 @description('Optional. Orchestration strategy: openai_function or semantic_kernel or langchain str. If you use a old version of turbo (0301), please select langchain. If the database type is PostgreSQL, set this to sementic_kernel.')
 @allowed([
   'openai_function'
@@ -210,6 +198,9 @@ param azureOpenAIEmbeddingModelVersion string = '2'
 @description('Optional. Azure OpenAI Embedding Model Capacity - See here for more info https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/quota .')
 param azureOpenAIEmbeddingModelCapacity int = 100
 
+@description('Optional. Azure Search vector field dimensions. Must match the embedding model dimensions. 1536 for text-embedding-ada-002, 3072 for text-embedding-3-large. See https://learn.microsoft.com/en-us/azure/search/cognitive-search-skill-azure-openai-embedding#supported-dimensions-by-modelname.(Only for databaseType=CosmosDB)')
+param azureSearchDimensions string = '1536'
+
 @description('Optional. Name of Computer Vision Resource (if useAdvancedImageProcessing=true).')
 var computerVisionName string = 'cv-${solutionSuffix}'
 
@@ -858,7 +849,7 @@ module pgSqlDelayScript 'br/public:avm/res/resources/deployment-script:0.5.1' =
     tags: tags
     kind: 'AzurePowerShell'
     enableTelemetry: enableTelemetry
-    scriptContent: 'start-sleep -Seconds 300'
+    scriptContent: 'start-sleep -Seconds 600'
     azPowerShellVersion: '11.0'
     timeout: 'PT15M'
     cleanupPreference: 'Always'
@@ -1285,6 +1276,7 @@ module web 'modules/app/web.bicep' = {
         MANAGED_IDENTITY_RESOURCE_ID: managedIdentityModule.outputs.resourceId
         AZURE_CLIENT_ID: managedIdentityModule.outputs.clientId // Required so LangChain AzureSearch vector store authenticates with this user-assigned managed identity
         APP_ENV: appEnvironment
+        AZURE_SEARCH_DIMENSIONS: azureSearchDimensions
       },
       databaseType == 'CosmosDB'
         ? {
@@ -1383,6 +1375,7 @@ module adminweb 'modules/app/adminweb.bicep' = {
         MANAGED_IDENTITY_CLIENT_ID: managedIdentityModule.outputs.clientId
         MANAGED_IDENTITY_RESOURCE_ID: managedIdentityModule.outputs.resourceId
         APP_ENV: appEnvironment
+        AZURE_SEARCH_DIMENSIONS: azureSearchDimensions
       },
       databaseType == 'CosmosDB'
         ? {
@@ -1482,6 +1475,7 @@ module function 'modules/app/function.bicep' = {
         AZURE_CLIENT_ID: managedIdentityModule.outputs.clientId // Required so LangChain AzureSearch vector store authenticates with this user-assigned managed identity
         APP_ENV: appEnvironment
         BACKEND_URL: backendUrl
+        AZURE_SEARCH_DIMENSIONS: azureSearchDimensions
       },
       databaseType == 'CosmosDB'
         ? {