How to Use Google Gemini: App, API, and AI Studio (2026)

The short answer (August 2026): Google Gemini is a family of AI models available three ways: as a consumer chat app at gemini.google.com (free with compute-based limits), as a browser-based developer sandbox called Google AI Studio (free for testing), and as a direct paid API starting at $0.25 per million input tokens. The free app tier works for most everyday tasks; paid plans start at $7.99 per month (Google AI Plus) and reach $100 or more per month for AI Ultra, which adds higher limits, YouTube Premium, and background-agent features. Developers can generate a free API key through Google AI Studio without entering payment information.

Last verified: August 9, 2026, against official docs and pricing pages.

Three Paths Into Gemini

Google structures Gemini access around three separate products.

The Gemini app lives at gemini.google.com and is aimed at general users and knowledge workers. Sign in with any Google account. The free tier includes Gemini with compute-based limits that reset every five hours until a weekly cap is reached, per Google's I/O 2026 subscription announcement. On Android, you can also summon the Gemini overlay by long-pressing the Side button; iOS users access it inside the Google app.

Google AI Studio (aistudio.google.com) is a browser-based workbench aimed at developers. It lets you run prompts against any current Gemini model, inspect real-time token counts, stream output, and generate API keys. Studio usage itself is free; you add billing only when you deploy API calls from your own application.

The Gemini API is for programmatic access. Authenticate with an API key, choose a model by its ID, and pay per token on the paid tier. A free tier with rate limits exists for every model. The API is the path for building Gemini into your own products.

The Gemini App: Free vs. Paid Plans

AI Plus is $7.99/month in the US with a discount for new subscribers, per the official AI Plus announcement. AI Ultra starts at $100/month (5x limits over Pro) with a $200/month tier (20x limits over Pro). The USD price for AI Pro was not confirmed on official English-language pages checked August 9, 2026; visit one.google.com/about/plans for current pricing.

If you are an existing Google One Premium 2TB subscriber in an eligible region, you automatically receive Google AI Plus benefits at no additional charge.

Four Features to Try in the App First

Deep Research. Click "Deep Research" in the prompt toolbar or Tools menu. Gemini conducts a multi-step web research pass and generates a report with inline citations. It can also connect to your Gmail, Drive, and Google Chat to pull your own documents into the analysis. This is the most differentiated feature compared to a plain chat interface.

Gems. Gems are saved custom personas. Open the Gems manager on desktop, write instructions, give it a name, and save. A "code reviewer" or "email drafter" Gem you configure once is available in every future session without re-prompting. Gems are available on all paid tiers.

File and media uploads. Tap the plus icon next to the prompt field to attach images, PDFs, or Drive documents. Gemini 3.6 Flash supports 1 million tokens of context, so large documents load cleanly without chunking.

Gemini Spark (AI Ultra only, US only). Spark is a background agent that handles recurring tasks without you opening the app. It can parse monthly statements, extract school or work deadlines from email, synthesize meeting notes, and create polished documents from scattered sources. It runs even when your device is locked. Availability is US-only as of August 2026.

Google AI Studio: Get an API Key in Five Minutes

Google AI Studio is the fastest way to evaluate a specific Gemini model before writing any application code.

1. Go to aistudio.google.com and sign in with your Google account.
2. Click "Get API key" in the left sidebar. A key is created tied to your Google Cloud project at no cost.
3. Open a new prompt, paste your text or upload a file, and select a model from the dropdown. Token counts appear in real time.
4. Click "Get code" to export a working Python, JavaScript, or cURL snippet with your key pre-inserted.

You do not need a billing account to generate a key or to use Studio within free-tier rate limits. Add billing when you exceed those limits or move to production. Once you add billing, you choose Prepay (minimum $10 credit purchase, credits expire after 12 months) or Postpay (charged at month end). Starting March 23, 2026, new AI Studio users may be required to start on the Prepaid plan.

Gemini API Models and Pricing

All models have a free tier with rate limits and a paid tier billed per token. Verified prices from the official pricing page:

Gemini 3.6 Flash is generally available. It supports a 1 million token context window and scored 58.7% on SWE-Bench Pro and 83.0% on OSWorld-Verified in benchmarks, per Google DeepMind. The batch API cuts every listed price by 50% on both input and output. Gemini 3.x models include 5,000 free monthly Search-as-a-tool requests on the paid tier.

For most new API projects, start with Gemini 3.5 Flash-Lite for cost control and switch to Gemini 3.6 Flash when you need stronger reasoning or multimodal handling. Gemini 3.1 Pro Preview is the right call for complex or creative tasks that justify the higher cost, but note the input price doubles for prompts exceeding 200,000 tokens.

What Gemini Struggles With

The free app tier hits its compute limits during extended sessions, which can interrupt long tasks. If you rely on Gemini daily, the $7.99 AI Plus plan is the minimal investment to get reliable headroom.

Gemini Spark and the Daily Brief agent are US-only in August 2026. If you are outside the US, those features do not appear regardless of plan tier.

API free-tier rate limits are low enough that a prototype hitting them often is better served by adding a Prepay billing account than by optimizing prompts around the limits.

Gemini 3.1 Pro Preview's input pricing doubles above 200,000 tokens. If you are building long-context applications, benchmark your actual cost at that threshold before committing to it in production.

FAQ

Is Google Gemini the same thing as Google Bard?
Bard was the brand name Google used for its AI chat product until February 2024, when it was renamed Gemini. The underlying models have advanced significantly since that rebrand.

Can I use the free tier indefinitely without paying?
Yes. The free tier at gemini.google.com has no expiration. Limits reset every five hours, so most everyday tasks complete without hitting a cap. You run into limits primarily during extended coding or large document analysis sessions.

What is the difference between Gemini the app and Gemini the model?
"Gemini" names both the family of AI models Google builds and the consumer chat application. The app runs on the latest Gemini models; the API gives you direct access to specific model versions by model ID, so your application calls a predictable version rather than whatever the app defaults to.

Do I need to set up Google Cloud to use the Gemini API?
API keys are linked to a Google Cloud project, but you can create one free through AI Studio without separately navigating the Google Cloud console. Billing is managed through Google Cloud once you exceed free-tier rate limits.

Will my API prompts be used to train Google's models?
Prompts submitted through the Gemini API on a paid billing account are not used to train Google's models by default. Free-tier and consumer app usage may be reviewed by human raters; you can review and change those settings in your Google account's AI activity controls.

Stay current on Gemini and every other AI tool that matters to practitioners: get AI Weekly free, 3 issues a week, read by 40,000+ practitioners.