`langchain-google-genai` uses older `google-ai-generativelanguage`, not the newer `google-genai`

[t-dsai]

Hi,
My setup is based on Python 3.13, with the following libraries

google-genai==1.19.0
langchain==0.3.25
langchain-core==0.3.64
langchain-community==0.3.24
langchain-google-genai==2.1.5
langchain-google-community==2.0.7

I tried using langchain-google-genai library as follows:

from langchain_google_genai import ChatGoogleGenerativeAI
llm = ChatGoogleGenerativeAI(
    model="gemini-2.0-flash",
    api_key=my_api_key,
    temperature=0.5,
    )
result = llm.invoke("Sing a ballad of LangChain.")
print(result.content)

However, it threw me an error, with google.ai module not found.

Upon going through chat_models.py in langchain_google_genai directory, I found usage of the following library.
google.ai.generativelanguage_v1beta

Upon installing google-ai-generativelanguage, the issue was resolved. However, Google recommends migrating from old Gemini SDK to the new Gemini SDK [https://ai.google.dev/gemini-api/docs/migrate]

With the naming langchain-google-genai, it hints that the underlying library is using the new Gemini SDK.

Is it planned to move to new Gemini SDK soon? If not, would it be a good idea to highlight this requirement somewhere prominently.

My apologies if I overlooked/missed this requirement of langchain-google-genai.

[windkit]

S. Tomar (@t-dsai) I have tried to reproduce your issue with the following requirements.txt and install with pip install -r requirements.txt

google-genai==1.19.0
langchain==0.3.25
langchain-core==0.3.64
langchain-community==0.3.24
langchain-google-genai==2.1.5
langchain-google-community==2.0.7

and the code you posted, but it is working fine.

Please note that google-ai-generativelanguage is NOT the old Gen AI SDK google is deprecating, this package is the API package. The old Gen AI SDK is google-generativeai

Please try to fix the installation with pip install langchain-google-genai==2.1.5, if you are still seeing the errors, it would be great if you can post the stacktrace displayed with the error and also the list of libraries installed with pip freeze