Python: Phase 2: Embedding clients for Ollama, Bedrock, and Azure AI Inference

[eavanvalkenburg]

Summary

Add embedding client implementations to existing provider packages as part of the vector store & embeddings port.

New embedding clients

• OllamaEmbeddingClient — Text embeddings via Ollama's embed API (packages/ollama/)
• BedrockEmbeddingClient — Text embeddings via Amazon Titan on Bedrock (packages/bedrock/)
• AzureAIInferenceEmbeddingClient — Text and image embeddings via Azure AI Inference (packages/azure-ai/)
  • Accepts Content | str input — dispatches text to EmbeddingsClient and images to ImageEmbeddingsClient
  • Separate model IDs for text (AZURE_AI_INFERENCE_EMBEDDING_MODEL_ID) and image (AZURE_AI_INFERENCE_IMAGE_EMBEDDING_MODEL_ID)
  • Async context manager support for proper resource cleanup

Additional changes

• Rename EmbeddingCoT → EmbeddingT, EmbeddingOptionsCoT → EmbeddingOptionsT in core
• Add otel_provider_name passthrough to all embedding clients (OpenAI, Azure OpenAI, Ollama, Bedrock, Azure AI Inference)
• Register integration pytest marker in all packages
• Add lazy-loading namespace exports for Ollama and Bedrock embeddings in core
• Add azure-ai-inference dependency to azure-ai package
• Add image embedding sample using Cohere-embed-v3-english

Testing

• Unit tests for all three new clients (Ollama: 7, Bedrock: 7, Azure AI Inference: 17)
• Integration tests with three-marker pattern (@flaky, @integration, @skip_if_...)

Fixes #4164
Part of #1188

[markwallace-microsoft]

Python Test Coverage Report •

File	Stmts	Miss	Cover	Missing
packages/azure-ai/agent_framework_azure_ai
_embedding_client.py	113	12	89%	152, 167–170, 174, 178, 228, 238, 249, 258, 284
packages/core/agent_framework
_clients.py	85	3	96%	298, 491, 493
_types.py	1024	86	91%	59, 68–69, 123, 128, 147, 149, 153, 157, 159, 161, 163, 181, 185, 211, 233, 238, 243, 247, 273, 277, 632–633, 1004, 1066, 1083, 1101, 1106, 1124, 1134, 1151–1152, 1154, 1172–1173, 1175, 1182–1183, 1185, 1220, 1231–1232, 1234, 1272, 1499, 1551, 1642–1647, 1669, 1674, 1840, 1852, 2104, 2125, 2220, 2449, 2656, 2726, 2738, 2756, 2954–2956, 2959–2961, 2965, 2970, 2974, 3058–3060, 3089, 3143, 3162–3163, 3166–3170, 3176
observability.py	658	81	87%	360, 362–364, 367–369, 374–375, 381–382, 388–389, 396, 398–400, 403–405, 410–411, 417–418, 424–425, 432, 470, 561, 703, 706, 714–715, 718–721, 723, 726–728, 731–732, 760, 762, 773–775, 777–779, 783, 791, 892, 894, 1043, 1045, 1049–1054, 1056, 1059–1063, 1065, 1174–1175, 1177, 1335, 1434, 1604, 1607, 1666, 1834, 1988, 1990
packages/core/agent_framework/azure
_embedding_client.py	21	0	100%
packages/core/agent_framework/openai
_embedding_client.py	55	1	98%	97
TOTAL	21920	2776	87%

Python Unit Test Overview

Tests	Skipped	Failures	Errors	Time
4592	247 💤	0 ❌	0 🔥	1m 15s ⏱️

[Copilot]

Pull request overview

This PR adds embedding client implementations for Ollama, Bedrock, and Azure AI Inference as part of Phase 2 of the vector stores & embeddings port. It includes type variable renames, otel_provider_name passthrough additions to existing clients, pytest marker registration across packages, and lazy-loading namespace exports.

Changes:

• Three new embedding clients: OllamaEmbeddingClient, BedrockEmbeddingClient, AzureAIInferenceEmbeddingClient with full test coverage
• Type variable renames: EmbeddingCoT → EmbeddingT, EmbeddingOptionsCoT → EmbeddingOptionsT across core abstractions
• Added otel_provider_name parameter to OpenAIEmbeddingClient and AzureOpenAIEmbeddingClient for telemetry customization
• Registered integration pytest marker in all package pyproject.toml files for consistent test organization
• Added azure-ai-inference>=1.0.0b9 dependency to azure-ai package

Reviewed changes

Copilot reviewed 40 out of 42 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
python/packages/ollama/agent_framework_ollama/_embedding_client.py	New OllamaEmbeddingClient using Ollama's embed API with truncate/dimensions options
python/packages/bedrock/agent_framework_bedrock/_embedding_client.py	New BedrockEmbeddingClient using Amazon Titan Embeddings via invoke_model API
python/packages/azure-ai/agent_framework_azure_ai/_embedding_client.py	New AzureAIInferenceEmbeddingClient supporting both text and image embeddings with Content dispatching
python/packages/core/agent_framework/_clients.py	Type variable renames EmbeddingCoT → EmbeddingT for consistency
python/packages/core/agent_framework/observability.py	Type variable renames EmbeddingOptionsCoT → EmbeddingOptionsT
python/packages/core/agent_framework/openai/_embedding_client.py	Added otel_provider_name parameter passthrough
python/packages/core/agent_framework/azure/_embedding_client.py	Added otel_provider_name parameter passthrough
python/packages/*/pyproject.toml (multiple)	Added integration pytest marker registration across 18 packages
python/packages/core/agent_framework/ollama/__init__.py	Added lazy-loading exports for OllamaEmbeddingClient and related types
python/packages/core/agent_framework/amazon/__init__.py	Added lazy-loading exports for BedrockEmbeddingClient and related types
python/samples/02-agents/embeddings/azure_ai_inference_embeddings.py	Sample demonstrating image and mixed text+image embedding generation
python/uv.lock	Dependency updates including azure-ai-inference and greenlet s390x removals