Gemini Pricing in 2026: Plans, API Costs, and Is It Free?

Google Gemini is free to start, and for a lot of people that is the end of the story. If you want more, the paid consumer plans run $4.99 a month for Google AI Plus, $19.99 for Google AI Pro, and $99.99 to $199.99 for Google AI Ultra. Build with the API instead and you pay per token, starting at $0.25 per million input tokens for a new project.

The catch is that Google reshuffled and repriced these plans more than once in 2026, so a lot of the Gemini pricing guides you will find quote numbers that no longer exist. Below is every current price, reconciled from each vendor's own pricing page, plus the question the numbers exist to answer: which plan, if any, you should actually pay for.

How Much Does Gemini Cost? Every Plan at a Glance

Here is the whole consumer lineup in one place, at US prices as of October 17, 2026.

Two things cause most of the confusion.

Gemini is sold two completely different ways. One is the consumer app, where you pay a flat monthly subscription, which is the table above. The other is the developer API, where your software pays per token for whatever it sends and receives, with no subscription at all. Most people reading a pricing page want the first one. If you are wiring Gemini into your own product, skip down to the API pricing. There is also a third path for organizations: Gemini is sold per seat inside Google Workspace business and enterprise plans, which is a separate price book from the consumer tiers here.

The second point: the paid plans are bundled into Google One, the same subscription that sells cloud storage. That is why each tier comes with a big storage allowance attached, and it is also why canceling trips people up. We come back to that at the end. All the consumer plans bill monthly, with no meaningful annual-prepay discount, so the monthly figure is the figure. For the bigger picture of what these models actually are, our guide to what Google Gemini is covers the lineup in plain terms.

Is Gemini Free? What the Free Tier Actually Gives You

Yes. Gemini has a genuinely useful free tier, and for most people it is enough.

On the free plan you get the Gemini app at no cost, with Auto model selection as the default since October 9, 2026: most prompts go to Google's Flash-Lite model, and harder ones may go to Flash or Pro. You also get image generation, voice mode, 15 GB of shared Google storage, and Deep Research. The honest catch is throttling. Limits are compute-based and refresh every five hours up to a weekly cap, and during busy periods free users can lose compute-heavy features like Deep Research.

What that cap actually means is the part most pricing guides skip, and it is the number that decides whether you need to pay. Google does not publish an exact figure, and the caps shift, but it does publish the ladder: Plus gets 2x the free limits, Pro 4x, Ultra 5x or 20x Pro. Ask a few questions a day and you will never hit it. Run document-heavy research all day and you will hit it fast, which is the real signal to consider a paid plan.

The other free-tier ceiling is the one almost every comparison gets wrong: the context window is gated by plan, not shared across them. Per Google's own limits page, the Gemini app gives you 32K tokens with no AI plan, 128K on Google AI Plus, and 1 million on AI Pro and AI Ultra. Google's own yardstick puts 1 million tokens at about 1,500 pages of text; the free tier's 32K is roughly 50. So the headline million-token figure you see quoted for Gemini is a paid feature, and if your reason for considering a paid plan is dropping whole documents into one prompt, that is the number that decides it, not the usage cap.

There is also a free tier on the developer side. The Gemini API gives you free, rate-limited access to a subset of models through Google AI Studio with no credit card; Gemini 4 Argon and 3.1 Pro are not on it. The trade is that, outside the EEA, Switzerland and the UK, free-tier traffic can be used to improve Google's products and human reviewers may read it; inside those regions, Google's terms apply the paid-services data terms to unpaid API quota as well. Either way, treat it as a prototyping tier and keep anything confidential out of it. Paid API usage, by contrast, is not used to train Google's models. More on that in the API section below.

Google AI Plus ($4.99 a Month): The Cheap Entry Tier

Google AI Plus is the budget paid plan, and it just got cheaper. In June 2026, Google cut Plus from $7.99 to $4.99 a month and doubled its storage from 200 GB to 400 GB, part of a broader price war among AI subscriptions.

For $4.99 you get roughly double the free-tier usage limits, a 128K context window (up from 32K), the Flash model, Gemini Omni Flash for video generation, 200 Google Flow credits, and the 400 GB of storage. It stops short of Gemini 4 Argon, which the next tier adds.

Who it is for: light users who keep bumping into the free-tier caps but cannot justify $20 a month. At $4.99 it is one of the cheapest serious AI subscriptions from a major lab, and the storage alone makes it competitive with a plain Google One plan.

Google AI Pro ($19.99 a Month): The Mainstream Plan

If you have heard of "Gemini Advanced," this is its descendant, though the naming took a detour that trips people up at checkout. Gemini Advanced became Google One AI Premium, and Google One's own plan FAQ now says the AI Premium plan is called Google AI Plus: the $4.99 tier above, with 400 GB of storage. The full-strength tier that most people mean by "Gemini Advanced" is this one, Google AI Pro at $19.99. So if you are following an old guide, check the storage figure and the price rather than the name: 400 GB means you are looking at Plus, 5 TB means Pro.

For that you get access to Gemini 4 Argon and Gemini 3.1 Pro, the full one-million-token context window, full Deep Research, video generation with Omni Flash, 1,000 Google Flow credits for Google's creative tools, access to Tasks in Gemini, formerly Spark (the agent, not offered in the EEA, UK, Switzerland or Nigeria), and roughly four times the free-tier usage limits. Storage jumps to 5 TB, up from the 2 TB it carried earlier in the year at the same price. The plan also brings Gemini into Gmail and Docs, raises the limits in Google's coding tools, and adds the Jules coding agent, an upgraded Gemini Notebook (formerly NotebookLM), and Antigravity, Google's agentic development environment.

The number that matters for comparison: $19.99 lands a cent under ChatGPT Plus at $20, so the two are a direct fight. Which one wins depends on whether you live in Google's apps or OpenAI's, and we break that down in Gemini vs ChatGPT. One thing worth checking before you pay: Google's student offer is live. The US student page currently offers eligible US college students (18+) Google AI Pro free for a year, including the 5 TB of storage and the higher Gemini limits. It can be redeemed until December 31, 2026, then renews automatically at $19.99 a month, and eligibility is re-verified each year. Google's student pages for Germany, France and Spain offer a year of Google AI Plus instead.

Who it is for: daily users who hit the free caps, anyone who wants Deep Research and video generation with more headroom, and people already paying for 2 TB of Google One who can fold storage and AI into one bill.

Google AI Ultra ($99.99 to $199.99 a Month): The Power Tier

Ultra is where the pricing gets genuinely confusing, because there are now two of them.

At Google I/O in May 2026, Google restructured the top of the lineup. It introduced a new $99.99 Ultra tier aimed at developers and serious creators, and it cut the existing top tier from $249.99 to $199.99. So the two Ultra prices are not a typo. They are two different plans.

Both Ultra tiers share the same headline features: Gemini 4 Argon and Project Genie, the interactive-world-model tool, are included on both the $99.99 and the $199.99 plan, as is 4K image and video upscaling, not gated to the top tier alone. The $99.99 tier gives you about five times the Pro plan's usage limits, higher Antigravity rate limits, 20 TB of storage, and a YouTube Premium plan. The $199.99 tier pushes usage limits to roughly twenty times Pro, up to 25,000 Google Flow credits, and 30 TB of storage.

That earlier $249.99 price is the one that made consumers wince, and the cut plus the new mid-tier is Google's answer to it. Even so, Ultra is a professional tool, and almost nobody needs the $200 plan for everyday use. It is built for people generating video at volume or running heavy agentic workflows. If you are not sure you need it, you do not.

Who it is for: high-volume creators, developers leaning on Antigravity and the agent tooling, and teams that will genuinely use tens of thousands of credits a month.

Gemini API Pricing: Pay-Per-Token by Model

If you are building software on Gemini rather than chatting with it, you do not buy a plan at all. You pay per token, the chunks of text the model reads and writes, billed separately for input and output. You can reach the same models two ways: the Gemini Developer API, which is the simplest path, or Vertex AI on Google Cloud, which carries the same per-token rates but adds enterprise billing, data governance, and SLAs. Here are the current Standard pay-as-you-go rates for the models worth using as of October 17, 2026.

A few things to read into that table. Flash-Lite is the floor and Pro is the ceiling, and the gap is wide: Gemini 3.1 Pro costs eight times more per input token than 3.1 Flash-Lite, and the older 2.5 Flash-Lite is cheaper still, though Google now limits the 2.5 models to accounts that already used them. For classification, extraction, and routine summarizing, the cheap models are usually fine, and routing simple work to Flash-Lite is where most teams find their savings.

Output costs far more than input, usually four to eight times, so long-winded responses are where bills balloon. The Pro models also carry a context cliff: once a single prompt crosses 200,000 tokens, Gemini 3.1 Pro input roughly doubles to $4.00 and output climbs to $18.00 per million. Retrieval pipelines that stuff large documents into every call can quietly cross that line on every request.

Google's flagship is now Gemini 4 Argon, generally available since October 13, 2026. One model to watch: Gemini 3.5 Pro still has not shipped. It has missed several targeted launch dates after Google reportedly rebuilt the base model, and Google has confirmed no release date, no context window, and no pricing for it. It does not appear on the Vertex AI pricing page (renamed the Gemini Enterprise Agent Platform in April 2026) or in the Gemini API model list. Until it lands, 3.1 Pro Preview is the top Pro model in the general API. Two more changes worth flagging if you are pricing this from an older guide: Google shut down Gemini 2.0 Flash and 2.0 Flash-Lite on June 1, 2026 and deprecated 3.7 Flash and 3.5 Flash on October 8, 2026, so any table still listing them as live is out of date, and its newest Pro model and Argon have no free API tier, so experimenting with either costs money from the first call, though 2.5 Pro still carries a free tier. Most current models share a one-million-token context window, so the differences are about speed, reasoning quality, and price, not how much they can read. Google's developer API pricing page is the source of record, so check it before you commit a budget. For the developer's full cost reference, the Standard, Batch, Flex, and Priority service tiers, the free-tier and rate-limit ladder, and a worked example of a real API bill, see our Gemini API pricing guide.

The Hidden API Costs: Batch, Caching, Reasoning, and Media

The per-token rates are only half the API bill. Four things move the real number, and only the first two move it down.

Batch mode cuts every paid request in half. If a job does not need an instant answer, like overnight enrichment or bulk generation, sending it through the Batch API gets a flat 50% discount in exchange for up to 24 hours of latency. It is the single biggest saving most teams skip.

Context caching pays off when you reuse the same context. If every call ships the same long system prompt, policy document, or codebase, caching it means you pay roughly 10% of the input rate on the repeats, plus a small hourly storage charge. For chatbots and document tools that reread the same material, the savings are large.

Reasoning tokens are the one that surprises people. Gemini's thinking models bill their internal reasoning as output tokens, even when the visible answer is short. A model with a low headline rate can therefore cost more in practice than one with a higher rate but terser reasoning. When you compare API prices, the sticker rate is a starting point, not the bill.

Then there are the separate meters. Image and video generation are not priced in text tokens at all. Google's newest image model, Nano Banana 2.1, runs about $0.034 per 1K image and $0.113 at 4K, and the Gemini API's Veo 3.1 video generation ran from $0.05 a second on the Lite model up to $0.60 a second for Standard at 4K on Google's pricing page as archived on October 15, 2026; Google's deprecations page now shows the three Veo 3.1 API preview models as shut down on October 22, 2026, with Omni Flash as the recommended replacement, and Veo 3.1 remains listed on Google Cloud's Gemini Enterprise Agent Platform. Audio is metered too: audio input costs more per token than text, and the Live API voice models bill separately again. Grounding a response in live Google Search is its own line: the Gemini 3 family gets 5,000 grounded prompts a month free and then $14 per thousand, while the 2.5 models get 1,500 free requests a day before the rate rises to $35 per thousand.

Gemini vs ChatGPT vs Claude: Which Is Cheaper?

On the consumer side, Gemini is the price leader. Its $4.99 Plus tier undercuts everyone, and its $19.99 Pro plan matches ChatGPT Plus and Claude Pro, which both sit around $20. If you are choosing purely on price, Gemini wins the entry tier outright.

On the API the picture is more nuanced, and it is where the cheapest sticker price can mislead.

Gemini's Flash models are among the cheapest credible options for high-volume work. Gemini 3.8 Flash at $0.75 input and $3.75 output is a fraction of what the frontier models charge: GPT-6.1 Sol runs $2.00 and $10.00 (and GPT-6 Astra $10.00 and $50.00), and Claude Sonnet 5.5 $2.00 and $10.00. The fair fight is Flash against each rival's own cheap tier, and there Gemini no longer wins: GPT-6 Luna and Claude Haiku 5.5 (on prompts up to 100,000 tokens) both list at $0.10 input and $0.50 output, below any Gemini model open to new projects.

But remember the reasoning-token tax. All three providers charge for the model's internal thinking on their reasoning tiers, so a head-to-head on base rates can flip once you measure how verbose each model's reasoning actually is on your workload. The honest move is to benchmark your real prompts on two or three models before committing, not to pick the lowest number in a table.

For the full feature-by-feature picture rather than just price, see our comparisons of Gemini vs ChatGPT and Claude vs Gemini.

Which Gemini Plan Should You Actually Pay For?

Most people overbuy. Here is the honest decision, from the bottom up.

Stay free if you ask Gemini a handful of questions a day and rarely run Deep Research. The free tier covers it, and you will know you have outgrown it the day you keep hitting the five-hour cap mid-task.

Pay $4.99 for Plus if you keep hitting the free caps but do not need Argon, or if the 400 GB of storage alone is worth it next to a plain Google One plan.

Pay $19.99 for Pro if you use Gemini daily for real work, want Gemini 4 Argon and the Pro model with four times the free limits, or already pay for Google One storage and can consolidate. This is the right answer for most paying users.

Pay for Ultra only if you generate video at volume, lean on the agent tooling, or want the highest Antigravity rate limits. The $99.99 tier is the sane entry point; the $199.99 tier is for professionals who can name exactly why they need it.

Use the API instead of a plan if you are building a product. A subscription buys one person a seat in the app; the API meters whatever your software does, and for anything programmatic that is the only model that makes sense.

Two practical traps before you buy. First, because the plans are bundled into Google One, canceling Gemini and canceling your storage are not the same action, and people get billed after they think they quit. Cancel on the platform you subscribed through, whether that is the web, the Play Store, or the App Store, and confirm your Google One storage stays at the level you actually want. And if you start on a free trial, it converts to the full monthly price automatically, so set a reminder before it renews. Second, a paid plan does not buy identical features everywhere. Tasks in Gemini (formerly Spark) comes with Google AI Pro upward but not in the EEA, UK, Switzerland or Nigeria, so a buyer in those markets pays the same $19.99 or $199.99 for a thinner feature set.

Why Gemini's Price Tag Matters for Your Brand

Here is the part that outlasts any specific price. Whichever tier people pay for, Gemini is no longer just a chatbot they visit. It is the engine behind the AI answers in Google Search. AI Overviews and AI Mode run on Gemini and reach billions of people a month, which means the model is already summarizing your category and naming competitors to searchers who never open the app.

So the more useful budget question is not which plan you buy. It is whether Gemini mentions your brand at all when someone asks it what to use. That is the gap we help businesses close, and it does not come on a pricing tier.

The numbers above will move again, because Google has repriced Gemini repeatedly in 2026 alone, so confirm the current figure on Google's plans page or developer pricing page before you pay. And once you know what Gemini costs, the next question is what it says about you. Our Citation Interceptor maps where Gemini, Google AI Overviews, and the other AI engines cite sources your brand is missing from, so you can see which conversations to get into. You can learn the broader method in our guide to tracking AI visibility.

Frequently Asked Questions

How much does Gemini cost per month?

Gemini is free to start. Paid consumer plans are Google AI Plus at $4.99, Google AI Pro at $19.99, and Google AI Ultra at $99.99 or $199.99 a month, as of October 17, 2026. The developer API is separate and billed per token, starting at $0.25 per million input tokens for a new project.

Is Gemini free?

Yes. The free tier routes most prompts to Google's Flash-Lite model (harder ones may go to Flash or Pro) and gives you image and voice, Deep Research, and 15 GB of storage, with compute-based limits that refresh every five hours. The API also has a free, rate-limited tier, though outside the EEA, Switzerland and the UK, free-tier traffic can be used to improve Google's products.

Is Gemini Advanced the same as Google AI Pro?

Roughly, but the naming is messier than a straight rename. "Gemini Advanced" became Google One AI Premium. Google One's plan FAQ now says the AI Premium plan is called Google AI Plus, which is the $4.99 tier with 400 GB of storage, while Google AI Pro is the $19.99 tier with 5 TB. If an older guide quotes "Gemini Advanced pricing" at around $20 a month, it is describing today's Pro tier. Go by the price and the storage allowance rather than the plan name.

Is Gemini cheaper than ChatGPT or Claude?

On consumer plans, yes. AI Plus at $4.99 undercuts the roughly $20 ChatGPT Plus and Claude Pro tiers, and AI Pro matches them at $19.99. On the API, Gemini's Flash models undercut the rivals' mid tiers but not GPT-6 Luna or Claude Haiku 5.5, and reasoning-token charges mean you should benchmark your own workload before deciding.

Sources

• Gemini Developer API pricing - Google AI for Developers - ai.google.dev/gemini-api/docs/pricing
• Google AI plans - Google One - one.google.com/about/google-ai-plans
• Google AI Pro and Ultra subscriptions - Gemini - gemini.google/subscriptions
• Gemini Apps limits and feature availability - Gemini Apps Help - support.google.com/gemini/answer/16275805
• Gemini API release notes - Google AI for Developers - ai.google.dev/gemini-api/docs/changelog
• Introducing Gemini 4 Argon - blog.google - blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-argon
• API pricing - OpenAI - developers.openai.com/api/docs/pricing
• Pricing - Claude API Docs - platform.claude.com/docs/en/about-claude/pricing
• Everything new in our Google AI subscriptions, fresh from I/O 2026 - blog.google - blog.google/products-and-platforms/products/google-one/google-ai-subscriptions
• Google AI Plus gets price drop to $4.99 and storage bump - 9to5Google - 9to5google.com/2026/06/08/google-ai-plus-price-drop
• Google just fired a warning shot in the AI subscription price wars - TechCrunch - techcrunch.com/2026/06/09/google-just-fired-a-warning-shot-in-the-ai-subscription-price-wars