Google SDK LlamaIndex Implementation

[hristogg]

Hi,

I have been building RAG with LlamaIndex and GCP suite for some time and have the following confusion/question :)
I started out mixing the Vertex implementation with GenAI, however after some issue with quotas and following this article
https://cloud.google.com/vertex-ai/generative-ai/docs/migrate/migrate-google-ai
I figured the VertexSDK is the proper one for productoin use cases and moved solely to it.

However I am not sure I fully understand the llamaindex approach, as from what I see (and here is where I am not sure how correct I am) it seems like that GenAI SDK has more support and releases.
Especially concerning for me seems the note on llama-index-llms-vertex in pip

NOTE: Gemini has largely been replaced by Google GenAI. Visit the Google GenAI page for the latest examples and documentation.

Am I missing something? Which one is the way to go and would woud the Vertex one be depreciated?

[dosubot[bot]]

The recommended approach for using LlamaIndex with Google Cloud Platform is to use the llama-index-llms-google-genai package, which utilizes Google's latest unified SDK. The Vertex class in LlamaIndex is marked as deprecated, with a recommendation to switch to the Google GenAI SDK. This suggests that the Vertex SDK is likely to be deprecated in favor of Google GenAI ^[1][2].

Additionally, the Google GenAI SDK is recommended for scenarios requiring streaming chat functionality, as it supports streaming capabilities that are not implemented in the Vertex SDK ^[3].

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}

[hristogg]

If anyone else is wondering the same, I am ont sure but I think you can actually access Vertex through the google-genai package through # Set environment variables
!export GOOGLE_GENAI_USE_VERTEXAI=true
!export GOOGLE_CLOUD_PROJECT='your-project-id'
!export GOOGLE_CLOUD_LOCATION='us-central1'

[logan-markewich]

@hristogg this is in the docs as well
https://docs.llamaindex.ai/en/stable/examples/llm/google_genai/#vertex-ai-support