The ChatCompletionClient.load_component() method was failing with ValueError("Invalid") when trying to load OllamaChatCompletionClient using provider configuration. This prevented users from using the same configuration-based loading approach available for other chat completion clients.

Problem

The following code would fail with a ValueError: Invalid error:

import asyncio
from autogen_core.models import ChatCompletionClient

async def main():
    config = {
        "provider": "OllamaChatCompletionClient",
        "config": {
            "model": "qwen3",
            "host": "http://1.2.3.4:30130",
        }
    }
    client = ChatCompletionClient.load_component(config)

if __name__ == "__main__":
    asyncio.run(main())

While the direct instantiation approach worked fine:

from autogen_ext.models.ollama import OllamaChatCompletionClient

client = OllamaChatCompletionClient(
    model="qwen3", 
    host="http://1.2.3.4:30130"
)

Root Cause

The issue occurred because OllamaChatCompletionClient was missing from the WELL_KNOWN_PROVIDERS dictionary in _component_config.py. When load_component() couldn't find the provider in the known providers list, it attempted to use "OllamaChatCompletionClient" directly as a module path. Since this string contains no dots, the parsing logic provider.rsplit(".", maxsplit=1) would return a single-element list, causing the validation len(output) != 2 to fail and raise ValueError("Invalid").

Solution

Added the missing entry to WELL_KNOWN_PROVIDERS:

WELL_KNOWN_PROVIDERS = {
    "azure_openai_chat_completion_client": "autogen_ext.models.openai.AzureOpenAIChatCompletionClient",
    "AzureOpenAIChatCompletionClient": "autogen_ext.models.openai.AzureOpenAIChatCompletionClient",
    "openai_chat_completion_client": "autogen_ext.models.openai.OpenAIChatCompletionClient",
    "OpenAIChatCompletionClient": "autogen_ext.models.openai.OpenAIChatCompletionClient",
    "OllamaChatCompletionClient": "autogen_ext.models.ollama.OllamaChatCompletionClient",  # Added
}

Changes

• Core fix: Added OllamaChatCompletionClient mapping to WELL_KNOWN_PROVIDERS in _component_config.py
• Tests: Added test_ollama_in_well_known_providers() to validate the provider mapping
• Integration tests: Added test_ollama_load_component() and test_ollama_load_component_via_class() to test the load_component functionality

Impact

• ✅ The original issue's code example now works without errors
• ✅ OllamaChatCompletionClient can be loaded using the same configuration approach as other clients
• ✅ Maintains consistency with existing OpenAI and Azure OpenAI client patterns
• ✅ No breaking changes to existing functionality
• ✅ Full backward compatibility preserved

Fixes #6994.

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.