Google Gemini

By Google

Google's Gemini models via the AI Studio API — long context, multimodal, with a free tier

Website ↗ Docs ↗

APIProprietary

#llm #multimodal #long-context #free-tier

Free tier Paid

Doc version 1.0 · Updated June 28, 2026

Related concepts

Harness Engineering Doc version 1.0 · Updated June 28, 2026 Wrapping an LLM in scaffolding — a control loop, tools, a sandbox, guardrails, and evaluation — so it behaves reliably in production.

Related tools

LiteLLM

Gateways & Routers

Overview

Gemini is Google’s family of frontier models, reachable through the AI Studio API with a free tier that makes it an easy default for demos.
It offers very large context windows and native multimodal inputs (text, images, audio, video).

Common model ids:

gemini-2.5-flash — fast and cheap, generous free tier
gemini-2.5-pro — most capable, for harder reasoning

flowchart LR
  app[your app] --> g[Gemini API]
  g --> reply[reply]

The Code samples tab shows the native SDK and the LiteLLM route — pick from the selector to compare.

When to use it

Reach for Gemini when you want a free tier to prototype with, very long context, or built-in multimodal input — and route it through LiteLLM to stay portable.

Pricing

Free tier on Flash / Flash-Lite (rate-limited, no card). Paid prices per million tokens (input / output)

Plan	Price	Notes
Flash-Lite (free)	$0	Rate-limited, no card
Gemini 2.5 Flash-Lite	$0.10 / $0.40	Cheapest paid tier
Gemini 2.5 Pro	$1.25 / $10	Most capable

Source ↗ · as of 2026-06-27

Example

Python · generate

When to use

Reach for the native SDK when you want Gemini-specific features like multimodal inputs or the large context window.

How it flows

generate_content sends the prompt to the AI Studio API; the reply text is on resp.text.

Structure

flowchart LR
  app[your app] --> g[Gemini API]
  g --> reply[reply]

from google import genai

client = genai.Client()  # reads GEMINI_API_KEY

resp = client.models.generate_content(
    model="gemini-2.5-flash",
    contents="Summarize Gemini in one line.",
)
print(resp.text)

When to use

Route through LiteLLM to keep your agent provider-agnostic — swap models by changing one string.

How it flows

gemini/<model> tells LiteLLM to call Google AI Studio; the rest of the call is the OpenAI-style shape.

Structure

flowchart LR
  app[your app] --> lite[LiteLLM]
  lite -->|gemini/| g[Gemini API]

from litellm import completion

# The gemini/ prefix routes to Google AI Studio with your GEMINI_API_KEY.
resp = completion(
    model="gemini/gemini-2.5-flash",
    messages=[{"role": "user", "content": "Summarize Gemini in one line."}],
)
print(resp.choices[0].message.content)

Example

Google Gemini — a minimal generate call

A real, runnable mini-project. Download it and run with Docker.

View samples/gemini_1 on GitHub ↗

June 28, 2026

Google Gemini — a minimal generate call

A tiny Gemini script: it sends one prompt to a Gemini model with the google-genai SDK and prints the reply.

Configure

cd samples/gemini_1
cp .env.sample .env
# edit .env: set GEMINI_API_KEY (and optionally MODEL)

Get a key at aistudio.google.com (free tier). MODEL defaults to gemini-2.5-flash. .env is gitignored — only .env.sample is committed.

Run with Docker

cd samples/gemini_1
docker build -t aas-gemini .
docker run --rm --env-file .env aas-gemini "Summarize Gemini in one line."

Run with Docker (in a devcontainer with DooD)

In a dev container that talks to the host Docker daemon (Docker-outside-of-Docker), the foreground docker run above often prints nothing and exits 0 — but the run itself succeeds. The script runs to completion and Docker captures all of its output; only the live attached stream drops it over the VM boundary. You can confirm this: docker logs on the same container shows the full output, the container exits 0, and it is not an OOM. Run detached and follow the logs instead:

cd samples/gemini_1
docker build -t aas-gemini .
docker logs -f "$(docker run -d --env-file .env aas-gemini \
  "Summarize Gemini in one line.")"

Run locally

cd samples/gemini_1
pip install -r requirements.txt
python app.py "Summarize Gemini in one line."

python-dotenv loads .env automatically.

Example run

Output varies by model and run — LLMs are non-deterministic, so the wording differs each time. Below is one run with gemini-2.5-flash.

Google Gemini is Google's multimodal AI model designed to understand, process,
and generate various forms of information, including text, images, audio, and
video.

Files

.env.sample

# Copy this file to `.env` and add your Google AI Studio key.
#   cp .env.sample .env
#
# Get a key at https://aistudio.google.com (free tier available)
GEMINI_API_KEY=

# Model id — see https://ai.google.dev/gemini-api/docs/models
MODEL=gemini-2.5-flash

app.py

"""Google Gemini: a minimal generate call with the google-genai SDK.

Sends one prompt to a Gemini model and prints the reply.

Run:
    docker build -t aas-gemini .
    docker run --rm --env-file .env aas-gemini "Summarize Gemini in one line."
"""

import os
import sys

from dotenv import load_dotenv
from google import genai

# Load .env when running locally; in Docker the key comes from --env-file.
load_dotenv()

client = genai.Client()  # reads GEMINI_API_KEY
MODEL = os.environ.get("MODEL", "gemini-2.5-flash")


def main() -> None:
    prompt = " ".join(sys.argv[1:]) or "Summarize what Google Gemini offers in one sentence."

    resp = client.models.generate_content(model=MODEL, contents=prompt)
    print(resp.text)


if __name__ == "__main__":
    main()

Dockerfile

# Minimal image for the Google Gemini sample.
FROM python:3.13-slim

ENV PYTHONUNBUFFERED=1

WORKDIR /app

COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt

COPY app.py .

# Provide your config at run time (.env is NOT baked into the image):
#   docker run --rm --env-file .env aas-gemini "<prompt>"
ENTRYPOINT ["python", "app.py"]