Gemini API: Pricing, Keys & How to Call It

Google's Gemini API is the value play among the big three: Gemini 3.1 Pro at $2/$12 per million tokens undercuts OpenAI and Anthropic while staying competitive on agentic benchmarks, and the Flash tiers go far cheaper. It also has a genuinely usable free tier. This guide covers the full price table by tier, the free-tier limits, how to get a Google AI Studio key, and how to call it — directly or through a gateway with 300+ other models.

What the Gemini API is

The Gemini API is Google's programmatic access to its model family — Pro (top reasoning), Flash (the fast, cheap workhorse), and Flash-Lite (the budget tier) — all natively multimodal and long-context. Its pitch is value: comparable agentic and coding quality to the US frontier at a noticeably lower price, plus a usable free tier for getting started.

Gemini API pricing

Per 1M tokens, USD, June 2026. Gemini 3.1 Pro prompts over 200K tokens bill at a higher tier (~$4/$18). Cached input is heavily discounted.

The Gemini tier ladder

Gemini's range spans an 8× output-price gap from Flash-Lite to Pro — and the Flash tiers are where most of the value sits:

The free tier & limits

Unlike the Claude and OpenAI APIs, Gemini has a genuinely usable free tier through Google AI Studio — rate-limited, but enough for prototyping and low-volume apps. The caveats: rate limits are tight, there's no uptime guarantee, and free-tier data-use terms differ from the paid tier, so don't send sensitive data on free. When you outgrow it, the paid tier raises limits and changes the data terms — a one-line change if you're already on the API.

How to get a Gemini API key

1. Go to Google AI Studio (aistudio.google.com) and sign in with a Google account.
2. Click Get API key and create one — the basic flow needs no Cloud project setup.
3. Store it as an env var: export GEMINI_API_KEY=...

How to call the Gemini API

Gemini offers an OpenAI-compatible endpoint, so the OpenAI SDK works directly:

from openai import OpenAI

client = OpenAI(
    base_url="https://generativelanguage.googleapis.com/v1beta/openai/",
    api_key="$GEMINI_API_KEY",
)

resp = client.chat.completions.create(
    model="gemini-3.1-pro",            # or gemini-3.5-flash, gemini-3.1-flash
    messages=[{"role": "user", "content": "Summarize this PDF and extract the tables..."}],
)
print(resp.choices[0].message.content)

The native Google GenAI SDK also works and exposes more multimodal features. For switching between providers, the gateway path keeps one code shape.

Pro vs Flash vs Lite

Calling Gemini through a gateway

To use Gemini alongside Claude, GPT and others under one key — with failover when Google is rate-limited — call it through a gateway. DataLLM Lab carries Gemini 3.1 Pro, 3.1 Flash, and 3.1 Flash-Lite today:

client = OpenAI(base_url="https://www.datallmlab.com/v1", api_key="$DATALLMLAB_API_KEY")
# reaches google/gemini-3.1-pro, anthropic/claude-opus-4.7, openai/gpt-5.4, ...

FAQ