This Add-on is using gemini 1.0 pro model for chatting.

1. Go to Google Ai and click the Get API key in Google AI studio
2. On API key page you can create new api key for gemini
   1. Note: if you are planing to use vertex api choose create new project
3. Copy the api key to addon settings
4. All done!

Note that: Vertex api is a SaaS (paid service) from Google so you will pay your own queries and you should know theese:

• Basic knowledge about Google Developer Console
• Basic knowledge about installing packages via terminal
• Basic knowledge about Google billing

1. create billing information
2. knowledge to gcloud-cli. ref: Install the gcloud CLI
3. Shortcut for Auth: If you test the vertex api from you account it will generate Auth credential. ref:
   • https://console.developers.google.com/apis/api/aiplatform.googleapis.com/overview?project=`PROJECT_ID`
   • if you create new project on Google Ai site you can learn your PROJECT_ID from Google Dev Console
4. After the installization of gcloud-cli you need to login via: gcloud auth login
5. Once you login via cli you need to follow theese steps (1-3) on your local: Vertex AI Gemini API Beginner Guide
6. Finally you can create JWT key by: gcloud auth application-default print-access-token on terminal
   • Note: You need to create manually JWT, Once you set the key to Gemini API key on addon settings, the addon will automaticly create new JWT's when token was expired

Vertex has internet access and it can be search on internet but Gemini API has no internet accsess and model update date is (06-2023) so it can not react to recent events. Basicly Gemini is free but outdated, Vertex is paid, outated but it can search on internet when info isn't in the model.

GCP > Cloud AI and Industry Solutions > Vertex AI Model Garden > Language > Gemini > Gemini Pro

Per Unit 1K output: 0.00036$

Per Unit 1K input: 0.00012$

• According the google Per unit means:

  • The SKU's pricing tier unit quantity. For example, if the tier price is $1 per 1000000 Bytes, then this column will show 1000000.

For reference: When I developed this addon, I sent 248 requests (it means 248 inputs and 248 outputs) and the cost was 0.019$.