Google API (Gemini models): pricing, free tier, rate limits and how to start

Connect Gemini API model choice to your application’s input, output and project controls. Use this hub to move from Google AI Studio setup to a small, measurable developer integration.

What Google offers through the API

The Gemini Developer API combines model access with Google AI Studio’s project and key management. The current quickstart uses the Google Gen AI SDK and the Interactions interface. Model entries cover text workflows and dedicated media capabilities, so begin with the kind of input and output your application requires. Official documentation.

For a document assistant, write an evaluation that separates reading the supplied document from supplying general background knowledge. A response can sound persuasive while missing the passage that actually answers the question. Save the evidence you expect the model to use and check whether the result preserves it. Add awkward examples such as unclear scans, incomplete tables and unrelated pages before you expand the workflow.

Google’s catalog distinguishes general Gemini models from image, speech, live-audio and other specialized offerings. Nano Banana names refer to documented image-model entries, whose API identifiers should be copied from that catalog. A capability available in one entry should not be assumed for every model carrying the Gemini family name. Official documentation.

Choose an evaluation artifact that matches the output. A generated image needs a visual review, an extracted field needs a known value, and a spoken response needs listening for intelligibility and timing. Do not collapse these into a single text-quality score. Make the input preparation reproducible so you can tell whether a change came from the model, a media conversion step or the instructions.

Gemini API keys belong to Cloud projects, and the current key documentation distinguishes authorization keys from older standard keys. Project access, billing and restrictions therefore belong in the setup checklist. A key that exists in a console is not sufficient evidence that every deployed environment can use it. Official documentation.

Keep a small operational record for each project: the application owner, approved input types, selected model identifier, billing contact and failure behavior. If the application receives private documents, decide how they are removed from debugging artifacts. An integration is easier to maintain when a developer can reproduce a failure without needing the original confidential file.

This hub links the technical and operational questions. The key guide follows the project setup path; Models explains how to narrow the catalog; Pricing keeps token and media charges separate. Rate limits covers capacity by project, while Errors helps distinguish a request-shape problem from access or service pressure. The Python tutorial keeps the first call small and prints diagnostics instead of hiding exceptions.

After the first successful call, preserve a baseline before adding conversation state, grounding or asynchronous work. Introduce each feature deliberately and record what changed in accepted results and usage. If the application needs a capability that the current interface does not provide, check the corresponding provider guide before substituting another interface. Similar naming is not proof that request fields or response handling are interchangeable.

Pricing at a glance

Read Gemini API pricing for token and media categories.

Models available

Explore the Google API model guide.

Free tier and how to get a key

Follow AI Studio key setup and inspect the model’s free-tier conditions.

Rate limits and tiers

Read Gemini project limits before attaching parallel workers.

Service status

Start in 5 minutes

Run the Google Gen AI Python tutorial.

result = client.interactions.create(model="gemini-3.8-flash", input="Hello", store=False)
print(result.output_text)

Recent changes

Use the AI API cost calculator to turn the model and workload you are considering into an estimate.

Last verified 12 Sep 2026 · Source ↗

Frequently asked questions

Sources

• Gemini API getting started ↗
• Gemini models ↗
• Developer API pricing ↗
• Rate limits ↗
• API-key management ↗
• API errors ↗
• Troubleshooting ↗
• Billing ↗
• Context caching ↗
• Batch API ↗
• Python SDK reference ↗

Last verified 12 Sep 2026 · Source ↗