Open
Description
Description
Currently, the chat API of the ModelProvider implementations is tested manually by connecting to a real LLM service. To improve test reliability and increase code coverage, we should implement a mock LLM service that simulates backend responses.
The mock service should be designed to validate the internal data mapping and transformation logic within the chat API. This will allow us to write deterministic unit tests that assert the correctness of request construction and response handling without relying on external services.
Version
No response