gRPC and ZScaler issue - langchain-google-vertexai

[ParinazMohsenpour]

I've encountered an issue while invoking the LLM using the langchain-google-vertexai integration. Based on initial investigation, the root cause appears to be related to gRPC proxy requirements and ZSclacer version which doesn’t handle http2 in the current environment.

I am currently looking for alternative options to langchain-google-vertexai and we’ve confirmed that the Google Vertex AI API (https://pypi.org/project/vertexai/) works correctly in the current networks settings and configurations. However, using google vertexai introduces certain limitations that require custom code to implement LangChain methods—specifically for LLM.invoke and llm.bind_tool.

Any recommended fixes or workarounds will be appreciated for this problem.