Google Gemini

Google Gemini is a foundational, general-purpose LLM. It is essentially the model that powers many of the AI services in Vertex AI and other Google products.

Python API

Google provides the Google Gen AI Python SDK for interacting with the Gemini developer API (as well as for interacting with Vertex AI)

You can install it via

pip install google-genai

and you can use it in a Python script via

from google import genai
from google.genai import types

To create a client using the Gemini API, you need a Gemini API key, which you then pass to a constructor function:

# replace the string with the actual key
client = genai.Client(api_key="GEMINI_API_KEY")

View available models

To see which models are available:

for model in client.models.list():
	print(model.name)

Uploading Documents

We might want to upload a document to a client. To do this, we can do the following:

document_file = client.files.upload(file='my_pdf.pdf')

And then if we wanted to, for example, summarize this document, we could do so via:

prompt = 'Summarize the following document'
 
cnfg = types.GenerateContentConfig(temperature=0.0) # setting a low temperature to preclude any creativity
 
response = client.models.generate_content(
	model = 'gemini-2.0-flash',
	config=cnfg,
	contents = [prompt, document_file]
)
 
print(response.text)

Further Reading

• Prompting Examples