Add Gemini Support

[hadirsa]

Hi everyone,

I’m proposing to add Gemini model support to the framework.
I’ve added the related files here:
hadirsa@9ec3b52

However, I haven’t been able to test it because the Vertex API is giving me a 403 error due to billing not being enabled on GCP.

Here’s the error:

{
  "@type": "type.googleapis.com/google.rpc.ErrorInfo",
  "reason": "BILLING_DISABLED",
  "domain": "googleapis.com",
  "metadata": {
    "containerInfo": "gen-lang-client-06884",
    "service": "aiplatform.googleapis.com",
    "consumer": "projects/gen-lang-client-068841"
  }
},
{
  "@type": "type.googleapis.com/google.rpc.LocalizedMessage",
  "locale": "en-US",
  "message": "This API method requires billing to be enabled. Please enable billing on project #gen-lang-client-06884 by visiting https://console.developers.google.com/billing/enable?project=gen-lang-client-068841 then retry. If you enabled billing for this project recently, wait a few minutes for the action to propagate to our systems and retry."
}

If someone can test this with a Vertex-enabled/billing-enabled GCP account, I’d appreciate it.
Once it’s confirmed working, I think we’re good to merge.

Thanks in a lot for any help!

[ddobrin]

Hi @hadirsa
have you configured your client as presented in this Google article?

Could you also update the models you are using, as Gemini 1.5 models are not available unless you have already used them (since April 29, 2025).
Gemini models

[donhuvy]

Current products and versions of Gemini are (June 25, 2025):

• Gemini Flash 2.0
• Gemini Pro
• Gemini with your search history.

[ddobrin]

Please see the current list of Gemini models available at this time, with Gemini Pro 2.5 and Flash 2.5 released

Gemini models

[ddobrin]

@hadirsa
Google has released the new Google Gen AI SDK, which supports both API Keys and credentials based auth. It is not yet supported in AI frameworks such as SpringAI and LangChain4j.

The Vertex SDK is deprecated and supported until June 24, 2026. That SDK is available within both Spring AI and LangChain4j.

I think it would be beneficial to have the Google Gen AI SDK supported in the above-mentioned frameworks first, to avoid duplication of effort in the integration with Embabel.

cc: @johnsonr

[heleihelei]

HI @ddobrin
Google is compatible with OpenAI, and Spring AI configuration parameters can be used。

Can Embabel use Spring Ai to support models compatible with OpenAI?

Google Doc: https://ai.google.dev/gemini-api/docs/openai

[ddobrin]

Hi @heleihelei
let me answer in 2 parts.

1. Yes, you can use Spring AI with the OpenAI module and connect to Gemini in VertexAI
   It was supported in the (now on the sunset path) Vertex AI Gemini module in Spring AI

2. What I would recommend is to use the new Google GenAI SDK integration in Spring AI, wich gives you both API Key (1) and credentials (2) for secure production deployments.

This new integration has been merged into Spring AI and will be released in the first milestone of Spring AI 1.1.0.

This is also the module I'd like to use for integrating Embabel with Gemini models.

[igordayen]

embabel-agent/embabel-agent-autoconfigure/models/embabel-agent-google-genai-autoconfigure

supported now:)

[igordayen]

@hadirsa - closing as implemented.