
Google AI Plus is $4.99, Pro $19.99 and Ultra $99.99 or $199.99. Here is what each tier buys, what changed in 2026, and what a real Gemini answer costs on a credit meter.
Google currently lists Google AI Plus with 400 GB of storage at $4.99 per month and Google AI Plus with 2 TB of storage at $9.99 per month. Google AI Pro is $19.99 per month. Google AI Ultra lists $99.99 monthly for 5x usage and $199.99 monthly for 20x usage compared with Pro. Gemini API usage is billed separately per token.

For the corresponding account, billing or archive decision, read ChatGPT + Claude + Gemini Cost: Monthly and Yearly, ChatGPT Business vs Gemini Workspace: Team Plan Guide, Move Gemini History to Claude: Takeout and Krater Import.
Google's U.S. pages list two Google AI Plus storage options: 400 GB at $4.99 per month on the Gemini plan page, and 2 TB at $9.99 per month on Google One. Google AI Pro is $19.99 monthly. Google AI Ultra lists a $99.99 monthly tier with 5x usage and a $199.99 monthly tier with 20x usage compared with Pro.
For a bundle decision rather than a model duel, Is Google AI Ultra Worth It? reviews Google AI Ultra by monthly value.
The individual Google AI Pro price is only one side of a buyer's choice; Google AI Pro versus ChatGPT Plus contrasts it with ChatGPT Plus.
If family access changes the math, our Gemini family plan guide explains which AI Pro benefits can be shared.
Workspace admins can compare the Gemini features bundled with Business Standard and Plus in Gemini Workspace seat pricing.
For the task-by-task model results behind this comparison, see Gemini vs ChatGPT.
Google AI plan features can include app access and Google services, but the exact bundle is regional and subject to change. Google's plan page also describes family sharing with up to five others for eligible plans. That is a Google account family-sharing feature, not an invitation to share one Gemini login or expose one another's chat history. See Google's U.S. plan comparison for the current included features.
The useful naming distinction is between the Gemini app, which people use for consumer chat and creation, and Google AI Plus, Pro, and Ultra, which describe paid subscription tiers on Google's U.S. plan page. Model names and subscription labels may change independently, so cite the specific plan name and country rather than writing only “Gemini Advanced.”
Google's Gemini plan page lists AI Plus with 400 GB at $4.99 per month, Pro at $19.99, and Ultra at $99.99 for 5x usage or $199.99 for 20x usage. Google One separately lists AI Plus with 2 TB at $9.99 per month.
Google AI Pro is listed at $19.99 per month in the U.S. The included product bundle is described on Google's live plan page and can change, so check it before subscribing for a feature such as higher model access, research tools, storage, or integrations. The listed price is a subscription charge, not a per-token allowance for the Gemini API.
For families, Google's plan comparison says eligible plans support sharing with up to five others. Family members should use their own Google accounts and check the current family-group rules, country requirements, and age restrictions. Sharing a subscription does not make consumer AI usage unlimited, and it does not pool Gemini API credits.
Google AI Ultra lists a $99.99 monthly tier with 5x usage and a $199.99 monthly tier with 20x usage compared with Pro.
Before moving from Pro to Ultra, list the specific features or higher usage allowance that you will use in a typical week. Compare those needs with the terms shown at checkout, not with a model name or a social post that may refer to a different date or market. The plan page is the authority for the currently available subscription level.
The Gemini API is billed separately from Google AI consumer subscriptions. Google's API pricing table uses U.S. dollars per million tokens. The rates below are the current rates read from the official page in Chrome on November 6, 2026; two promotional rates have scheduled increases, so the effective date matters. See the official Gemini API pricing table before shipping a budget.
| API model | Input per 1M | Output per 1M | Cached input / storage | Batch |
|---|---|---|---|---|
| Gemini 4 Argon | $2.00 through Nov. 30, 2026; $4.00 from Dec. 1 | $10.00 through Nov. 30; $20.00 from Dec. 1 | $0.10 through Nov. 30; $0.20 from Dec. 1. Implicit caching only. | 50% lower than standard where available. |
| Gemini 3.8 Flash | $0.75 through Dec. 31, 2026; $1.50 from Jan. 1, 2027 | $3.75 through Dec. 31; $7.50 from Jan. 1 | $0.075 through Dec. 31; $0.15 from Jan. 1, plus storage at $0.50 per 1M cached tokens per hour through Dec. 31 and $1.00 from Jan. 1. | 50% lower than standard where available. |
| Gemini 3.5 Flash-Lite | $0.30 | $2.50 | $0.03 cached input; storage is priced separately in the official table. | 50% lower than standard where available. |
Rates are not the entire application bill. Tool use, grounding, storage duration, image or audio input, and model-specific rules can add cost. Batch is a processing option with a 50% reduction where supported, not a universal discount on every request. Check the pricing table's model-specific notes before estimating.
Compare the products that actually match your workflow. Google AI Pro is listed at $19.99 per month in the U.S.; ChatGPT Plus is listed at $20 per month on OpenAI's U.S. plan page. The similar subscription price does not make the bundles identical: each service has its own model access, usage rules, storage, connected products, and regional terms.
Keep subscription billing separate from API billing. A consumer plan does not automatically include a fixed amount of API tokens, and an API key does not grant the app's subscription features. If you are evaluating an application, build a token budget from expected input and output volume and compare that with the API table. If you are choosing a personal workspace, compare the features you actually use and the cancellation and renewal terms.
Krater Pro provides a workspace for multiple model families rather than a separately priced Google AI subscription. The code-backed configuration lists Pro at $20 per month or $200 per year with 1,500 monthly credits. Its plan page describes 500+ models, including Gemini, Claude, and GPT, as well as Krater Agent. The selected model and task determine credit consumption, so there is no one fixed “price per Gemini answer” that applies to every prompt.
Use the credit balance and any in-product estimate for the specific model and workflow you intend to use. Do not assume that the Google AI Pro subscription or Gemini API rates are bundled into Krater, and do not describe Krater's credit system as a replacement for API billing. Read Krater's plan details and the official Gemini API pricing page as separate terms.
Stay on Free if you only need occasional answers. Pick Google AI Plus with 400 GB at $4.99, or the 2 TB version at $9.99, if you hit free limits but do not need the Pro model. Pick Google AI Pro at $19.99 if you want the flagship model, Deep Research and the storage bundle. Ultra only makes sense if you actually burn Flow credits or live in Agent Mode; Ultra is $99.99 per month for 5x usage or $199.99 for 20x usage compared with Pro. If you use the API, budget on January 2027 rates and watch thinking-token inflation on Pro models. If your usage spans several model families and you want a single bill, a per-answer plan like Krater's is the other shape worth pricing.
| Need | Route to evaluate | Budget question |
|---|---|---|
| Personal chat, writing, and app features | Gemini app with Free, Google AI Plus, Pro, or Ultra | Does the current plan include the feature and usage level your account needs? |
| Build Gemini into a product or automation | Gemini API | What are the expected input, output, cached tokens, storage hours, and tool calls? |
| Use Gemini alongside other model families | Krater Pro workspace | How many credits does the specific model and task consume? |
These routes solve different jobs. A consumer subscription gives access to the Gemini app and its current plan features. The API gives a developer token-metered access to supported models and options. Krater Pro is configured as a monthly credit workspace across 500+ models and includes Krater Agent; its credit use is not the same as a Google AI subscription or a Google API invoice.
Before selecting an API model, build an estimate from actual or expected traffic. If requests include long system instructions, repeated conversation history, images, or tool calls, include those in the estimate rather than counting only the user's last message. If you expect stable prefixes, compare the official cache terms and storage charges. If work can run asynchronously, check whether Batch is supported and whether the latency trade-off is acceptable.
For consumer plans, note the country, account type, price shown at checkout, and renewal date. Google's plan page may include introductory offers or family-sharing features with eligibility conditions. A screenshot from another country or a previous month is not a substitute for the live terms displayed to your account. For an API project, save the pricing table URL and check it again before every major release, especially around the published December 2026 and January 2027 rate changes.
A practical decision test is to write down one job for each route and run a small sample. Use the Gemini app for personal work, the API for a measured production workload, and a shared workspace only if the model variety and credit-based workflow fit the team. Keeping those budgets separate prevents a consumer subscription from being mistaken for included API credits.
Separate a consumer subscription budget from an API application budget. Google AI Plus, Pro, and Ultra are subscription plans for Google's products. Gemini API usage is metered by tokens and can include caching, storage, Batch processing, grounding, or other tool charges. A consumer plan should not be treated as prepaid API volume.
Here is a simple one-day API estimate using 1 million input tokens and 250,000 output tokens, with no cached requests or paid tools. At the Gemini 4 Argon promotional rates shown through November 30, 2026, input costs $2.00 and output costs $2.50, for $4.50 that day. At the rates shown from December 1, the same token volume costs $4.00 plus $5.00, or $9.00. For Gemini 3.8 Flash through December 31, 1 million input tokens cost $0.75 and 250,000 output tokens cost $0.9375, for $1.6875. At the published January 1 rates, that same workload costs $1.50 plus $1.875, or $3.375.
These examples deliberately hold model, input, output, and tool use constant. A real application may have more output than expected, may send the same system context on every request, may incur storage for cached content, or may use grounding. If you use caching, calculate both the cached-input price and any storage duration. The official table describes Gemini 4 Argon's caching as implicit only; do not budget an explicit cache creation workflow for that model unless the documentation changes.
Google's pricing table shows scheduled rate increases for Gemini 4 Argon on December 1, 2026 and Gemini 3.8 Flash on January 1, 2027. Put those dates into the application's budget review instead of reusing a pre-change quote. For Gemini 3.5 Flash-Lite, the table lists cached input separately and prices storage separately; verify that storage row directly before estimating a long-lived cache. Recheck Google's official API pricing page on the day you deploy or renew a budget.
The Gemini API table can show more than one rate for a model because an introductory price has an end date. Keep the model name, effective date, billing unit, and token type together in any budget. A row that says “$0.75 input” is incomplete unless it also says the observed Gemini 3.8 Flash rate applies through December 31, 2026 and changes on January 1, 2027.
Use separate columns for input, output, cached input, cache storage hours, and Batch eligibility. Then calculate from actual usage rather than applying the input rate to all tokens. If a request sends 900,000 input tokens and returns 100,000 output tokens, use 0.9 of the input price and 0.1 of the output price per million. Add storage and tool rows separately. This avoids undercounting output or forgetting that a cached prompt can still incur storage charges.
Keep a date-stamped copy of the official pricing URL in the project budget and revisit it before scheduled changes. If a model's name or version changes, do not carry over the old row automatically. Consumer subscriptions and API meters are separate products, so an unchanged Google AI app plan does not establish an API rate.
At the headline tier, no: Google AI Pro is $19.99 a month and ChatGPT Plus is $20. Google AI Plus is listed at $4.99 for 400 GB or $9.99 for 2 TB. ChatGPT has its own Free, Go, Plus, and Pro plans, so compare the features and storage you need rather than price alone.
It is worth it if you use Deep Research, the 1M context window, or the 5 TB storage and media credits bundled into Pro. If you only ask short questions a few times a day, Free or Google AI Plus at $4.99 for 400 GB may cover it; the 2 TB option is $9.99.
Free gives base model access with lower limits. Paid tiers raise usage limits and unlock the Pro model, larger context, media generation and storage. On the API, the free tier also trains on your content while the paid tier does not.
As a model, Pro-tier access moved off the free tier on 1 April 2026 per CloudZero's timeline. Students can get a free year of Google AI Pro through the offer on the subscription page; otherwise Pro costs $19.99 a month.
Google AI Plus with 400 GB at $4.99 a month is the lowest listed paid tier; Google One lists the 2 TB option at $9.99.
No. Consumer plans and the API are billed separately. API pricing is per million tokens on ai.google.dev, with promo rates that rise on 1 January 2027.
No. The Advanced name was retired in the 2026 rebrand. What was Gemini Advanced is now Google AI Pro at $19.99, and the top tier is Google AI Ultra.
Google's U.S. pages list AI Plus at $4.99 for 400 GB or $9.99 for 2 TB, Pro at $19.99, and Ultra at $99.99 for 5x usage or $199.99 for 20x usage compared with Pro. The real traps are not the sticker prices but the January 2027 API repricing and thinking tokens billed at output rates. Pick by which bundle extras you would pay for anyway.
More guides: ChatGPT pricing in 2026.
More guides: Grok pricing in 2026.