Gemini Pricing (2026): The $4.99 Plan Wins for Most People

Gemini costs $0 to $199.99 a month, but the $4.99 AI Plus tier is the best paid plan for most people. Stay free for ordinary chat, move to Pro for Gemini 4 Argon access, the 1-million-token context or a recurring need for more Flow capacity, and treat Ultra as a production-creator purchase rather than a better chatbot.

Gemini is Google's consumer AI app. Gemini pricing spans Free, AI Plus, AI Pro and two AI Ultra options, a separate developer API, and separate Google Workspace business plans. That is why two people can both say they "pay for Gemini" while buying completely different things.

Prices, versions, limits and benefits below were verified against Google's consumer plan page, Gemini subscriptions, Google support pages, and Developer API pricing on 14 October 2026. The recommendation turns on four measurable things: context length, Flow video output, storage value, and whether you need an app subscription, API usage, or Workspace seats.

Gemini pricing at a glance

The current US base consumer ladder is $0, $4.99, $19.99, $99.99, or $199.99 per month. Paid monthly for a full year, those four paid options total $59.88, $239.88, $1,199.88, and $2,399.88 respectively. Pro also supports annual billing; use the current checkout quote for its annual price. Google's plan page.

The table uses the base 400 GB Plus plan. Google also lists a 2 TB Plus configuration with extra Meet and Calendar benefits; check its displayed checkout price rather than assuming it costs $4.99. Google's Plus configurations.

The trap is treating these as five versions of the same chatbot. They are bundles. Storage, video credits, Google app access, and individual media perks account for much of the price jump. If those extras do not replace something you already buy or produce output you can value, they are not savings.

What the free Gemini plan really allows

The Free plan is not a throwaway demo. It now lists Flash-Lite access while keeping image generation and editing, Deep Research, Gemini Live, Canvas, Gems, Gemini Notebook, and 15 GB of storage. It does not include Argon. Google's subscription page.

Google's model-selection change began rolling out to Free users on October 9, 2026. Auto sends most prompts to Flash-Lite and may route prompts needing deeper reasoning to Flash or Pro; turning off smart model selection makes every response use Flash-Lite. That is different from choosing Pro yourself. Google's model-access changes. The free Gemini plan guide covers the remaining features and limits; the Gemini 3 review covers the capability side.

The honest limitation is that Google does not promise a fixed prompt count. Free has a 32,000-token context window, while prompt complexity, uploaded file size, conversation length, feature choice, and live capacity all affect how quickly the allowance is consumed. Context is the amount of text and other input the model can keep in one working session. A 32,000-token window is plenty for ordinary chat, but it is not the place to hand Gemini a large repository or a long stack of contracts at once.

Google now measures consumer usage by compute rather than a simple daily message counter. Access refreshes every 5 hours until a weekly cap is reached, simpler prompts consume less, and paid subscribers can continue with Flash-Lite after reaching their allowance. That makes the free tier useful but difficult to budget around if Gemini sits inside a deadline-driven workflow. Google's app limits.

Free Flow access is more concrete. With or without a subscription, users receive 50 free credits per day, and those credits do not roll over. At current Veo rates, one active day buys at most 5 Lite or 2 Fast generations for a non-Ultra user. A Quality generation costs 100 credits, so daily free credits cannot be saved for one. Omni Flash offers other resolutions and credit costs; paid monthly credits are an additional allowance. Google's Flow credits.

Who never needs to pay? Anyone using Gemini for ordinary questions, occasional Deep Research, short documents, light image work, and a few low-cost Flow experiments can remain on Free indefinitely. Pay only when the variable cap repeatedly blocks paid work, not because a plan page labels a tier "Pro."

Which paid Gemini plan is worth it

AI Plus is the default paid choice at $4.99

AI Plus includes Flash access, doubles the Free plan's Gemini app usage, expands context to 128,000 tokens, includes 400 GB of storage, and supplies 200 Flow credits plus 3,000 Flow Music credits each month. It also puts Gemini into Gmail and other Google apps.

That is enough for a solo operator who wants more dependable daily access, works across medium-size documents, and occasionally creates video. The annualized sticker price is $59.88. The recurring allowance is the limit: 200 monthly Flow credits can run out. Google's help pages disagree on whether Plus can buy extra AI credits, as explained below. Check your account's purchase option before treating an overage as either a small top-up or an automatic reason to move to Pro.

Skip Plus if Free rarely interrupts you. If you specifically need Argon in the app, Pro is the entry plan. Also skip Plus if your actual need is programmatic API access, because the subscription does not pay an API bill.

AI Pro earns $19.99 through capacity and bundling

AI Pro includes Gemini 4 Argon in the app, quadruples the Free plan's Gemini app usage, expands context to 1,000,000 tokens, includes 5 TB of storage, and supplies 1,000 Flow credits plus 10,000 Flow Music credits per month. Its bundle also includes $10 per month in Google Cloud credits, YouTube Premium Lite for one person, Google Home Premium Standard, and broader Gemini access across Google apps.

Argon access and the 1,000,000-token context are the clearest technical reasons to upgrade. A research-heavy founder, analyst, or technical operator can keep far more source material in one session. The 1,000 Flow credits are the clearest commercial reason: Pro adds 800 credits over Plus for $15 more, an incremental cost of 1.875 cents per credit.

Family sharing changes the math again. Google AI plans can be shared with up to five other people, making six people including the plan manager, although individual perks such as YouTube Premium are not shared. Flow credits and some product limits remain a shared pool, not a fresh allowance per relative. Pro costs about $3.33 per person across six people. At five people it is $4.00 each, below five separate Plus subscriptions at $4.99 each. At four people it is about $5.00 each, so five sharing users are the price break-even. Google's family-sharing rules.

Skip Pro if Argon access, the million-token context, storage, app integrations, or Flow capacity do not remove a measured constraint. Four times a variable allowance is still not a guaranteed prompt count.

AI Ultra is a production-volume purchase

AI Ultra 5x costs $99.99 per month. It raises Gemini app usage to 5x Pro, keeps the 1,000,000-token context, includes 20 TB of storage, 10,000 Flow credits, 30,000 Flow Music credits, $40 per month in Google Cloud credits, YouTube Premium for one person, and Google Home Premium Advanced.

AI Ultra 20x costs $199.99 per month. It raises app usage to 20x Pro, includes 30 TB of storage, 25,000 Flow credits, 30,000 Flow Music credits, and $100 per month in Google Cloud credits. Both Ultra options include Argon in the app.

Neither Ultra tier is a rational purchase for an ordinary chat user. They are for a creator or operator who can turn large video allowances, extreme usage headroom, or the bundled services into output worth more than $99.99 or $199.99 every month. If you cannot forecast that output before subscribing, the cheaper plan is the decision.

Flow credit math exposes the real break-even

Flow credits are where Gemini's plan names become actual production economics. One Veo 3.1 Lite generation costs 10 credits outside Ultra and 5 credits on Ultra. Fast costs 20 credits outside Ultra and 10 on Ultra. Quality costs 100 credits on every paid plan. A 10-second Gemini Omni Flash generation costs 15 credits at 720p or 7 credits at 360p. Ultra subscribers and family plan managers also have a zero-credit Lower Priority Veo 3.1 Lite option, which is unavailable to Ultra family members and non-Ultra users. Google's per-generation rates.

Using only the paid monthly allocations, and excluding daily credits, top-ups and the zero-credit Lower Priority option, Plus can produce at most 20 Lite, 10 Fast, 2 Quality, or 13 ten-second Omni generations at 720p, with 5 credits left in the Omni case. Pro can produce 100 Lite, 50 Fast, 10 Quality, or 66 ten-second Omni generations. Ultra 5x can produce 2,000 Lite, 1,000 Fast, 100 Quality, or 666 ten-second Omni generations. Ultra 20x can produce 5,000 Lite, 2,500 Fast, 250 Quality, or 1,666 ten-second Omni generations.

If the whole subscription price is assigned to Flow, one Lite generation costs roughly $0.25 on Plus, $0.20 on Pro, $0.05 on Ultra 5x, and $0.04 on Ultra 20x. One Quality generation works out to about $2.50, $2.00, $1.00, and $0.80 respectively. A ten-second Omni generation at 720p is about $0.38, $0.30, $0.15, and $0.12. This is deliberately conservative because it assigns no value to storage, Gemini chat, or the other bundle benefits.

Moving from Plus to Pro costs $15 more and adds 800 monthly credits, or 1.875 cents per additional credit. Compare that recurring capacity with any top-up offer your account actually displays before upgrading. Moving from Pro to Ultra 5x costs $80 more for 9,000 additional credits, or about 0.889 cents per added credit. Moving from Ultra 5x to Ultra 20x costs $100 more for 15,000 additional credits, or about 0.667 cents per added credit.

The hidden rules matter. Daily credits and paid monthly credits do not roll over. One request may return two generations, and Google charges credits for each generation. Paid plans include 1080p upscaling, while 4K upscaling is Ultra-only and costs 50 credits.

Google's Flow help page says Plus, Pro and Ultra can buy AI credits in supported regions except Japan. Its Plus guide and credit-purchase guide still say Plus cannot. That conflict means Plus top-ups are not a dependable budgeting assumption until your account confirms eligibility.

The purchase guide shows increments such as 2,500, 5,000 or 20,000 credits, but no public dollar quote. Purchased credits may expire under their purchase terms. Automatic reload is available to an eligible plan manager, with no more than five reload purchases per day, and a family draws from the shared pool. Compare the displayed purchase price with the next plan's monthly allowance.

Gemini API pricing is a separate bill

The Gemini Developer API charges for the tokens your software sends and receives. AI Plus, Pro, and Ultra are app subscriptions with their own benefits; subscribing does not turn metered API calls into unlimited app usage. Google's AI Studio plan rules exclude Argon from the Pro allowance inside AI Studio, even though Pro includes it in the consumer Gemini app. Ultra includes Argon in AI Studio; direct API calls still use separate billing. A token is a small piece of text the model reads or writes, and the text models' output rate includes thinking tokens, even when those tokens are not visible in the final answer.

Google's current pricing page lists Gemini 3.8 Flash and the previous-generation Gemini 3.6 Flash at $0.75 input and $3.75 output per 1 million tokens through December 31, 2026. On January 1, 2027, both rates double to $1.50 input and $7.50 output. Budget a recurring workload at the January rate before you commit to a customer price. Google's Gemini Developer API pricing.

Current text, Live, and TTS model prices

These prices are in USD per 1 million tokens, verified on October 14, 2026. Service is Standard unless a row names Batch, Flex or Priority. Argon rates change on December 1, 2026; the dated Flash and TTS rates change on January 1, 2027. Output prices include thinking tokens where applicable. Live means streaming conversation; TTS means text-to-speech, which turns written input into spoken audio. The free-tier column refers to input and output for the named service, not every paid feature.

Source for every row and dated rate: Google's API pricing page, last updated October 14, 2026. Gemini 3.6 Flash is the previous-generation model. The current page no longer lists Gemini 3.7 Flash or Gemini 3.5 Flash, so their former rates are not quoted here. Gemini 3.5 Flash-Lite is the cheaper baseline for high-volume extraction or classification; test whether it meets your task's quality bar before paying for Flash or Pro.

Since October 13, 2026, Gemini 4 Argon has been available through the paid Gemini API. Google now also lists Argon in the app for AI Pro and both Ultra options, with availability in select countries. Budget ongoing Argon API work at the December 1 rates above: app access does not cover those token charges.

What a monthly bill becomes on January 1

Suppose your product handles 10,000 interactions a month, each with 4,000 input tokens and 1,000 output tokens, including any thinking tokens. That is 40 million input tokens and 10 million output tokens. At Gemini 3.8 Flash or 3.6 Flash Standard rates, the token-only bill is:

• Through December 31, 2026: 40 × $0.75 input + 10 × $3.75 output = $30 + $37.50 = $67.50 per month.
• From January 1, 2027: 40 × $1.50 input + 10 × $7.50 output = $60 + $75 = $135.00 per month.

The same workload costs $67.50 more per month, a 100% increase, without serving another customer. This example excludes caching, grounding, media and other charges; it is calculated from Google's published token rates.

Per interaction, that is $0.00675 now and $0.0135 from January 1. The current 1,000-interaction totals are $6.75 on Standard Flash, $3.70 on 3.5 Flash-Lite, and $20.00 on 3.1 Pro Preview below its long-context threshold. At the current Flash rate, the $4.99 Plus sticker equals about 739 comparable API interactions and the $19.99 Pro sticker about 2,961. These are price comparisons, not subscription entitlements or an exchange of app fees for API tokens.

Batch, Flex, and cached-input charges

Batch processes work asynchronously, with a target turnaround of 24 hours and input/output prices at 50% of Standard. For 3.8 Flash and 3.6 Flash, Batch and Flex cost $0.375 input / $1.875 output now, then $0.75 / $3.75 from January 1, 2027. The example workload therefore costs $33.75 per month now and $67.50 after the increase when billed at those rates. Use Batch for work that can wait, such as overnight document extraction. Google's Batch guide, model-specific prices.

For 3.5 Flash-Lite, Batch and Flex cost $0.15 input / $1.25 output. For 3.1 Pro Preview they cost $1.00 / $6.00 at or below 200,000 prompt tokens and $2.00 / $9.00 above that threshold. Priority service on 3.8/3.6 Flash costs $1.35 / $6.75 now, rising to $2.70 / $13.50 on January 1. These service prices do not change what an app subscription buys. Google's service rows.

Caching lets you reuse input instead of paying the full input rate again. Read the cached-input fee and any hourly storage fee separately, and use the exact model and service row:

• Gemini 4 Argon: implicit caching only; explicit caching is not supported. Per 1M cached input tokens, Standard is $0.10 through November 30, 2026 and $0.20 starting December 1, 2026. Batch and Flex are each $0.05 then $0.10 on those dates; Priority is $0.18 then $0.36. Google lists no separate cache-storage price for Argon.
• 3.8 Flash and 3.6 Flash: Standard cached input is $0.075 per 1M tokens now, then $0.15 from January 1. Batch/Flex cached input is $0.0375 now, then $0.075. Storage is $0.50 per 1M cached tokens per hour now, then $1.00.
• 3.5 Flash-Lite: cached input is $0.03 on Standard and $0.02 on Batch/Flex. Storage is $1.00 per 1M tokens per hour.
• 3.1 Pro Preview: cached input is $0.20 for prompts at or below 200,000 tokens or $0.40 above, with $4.50 per 1M tokens per hour for storage. Its Batch/Flex cache rates are the same as Standard.

Those are the pricing page's caching rows. A cheaper token rate does not make keeping a large cache around free. Google's caching documentation also describes automatic implicit caching; savings depend on actual cache hits. Do not assume that a 50% Batch input/output discount halves every cache or storage fee.

Voice pricing uses its own meter

For 3.8 Live, 3.8 Live Extended Thinking and 3.1 Flash Live Preview, Google also expresses audio input as $0.005 per minute and audio output as $0.018 per minute; image/video input is $0.002 per minute. Input and output are separate charges, so a customer's elapsed call length is not automatically the amount of audio billed in each direction. The pricing page does not attach the Flash text January step-up to these Live rows. Google's Live rates.

Google's approximate combined rates are $0.0368 per minute for 3.5 Live Translate, $0.009 for 3.5 Transcribe Live, and $0.005 for 3.5 Transcribe. The underlying bill uses input/output tokens, and those minute estimates depend on Google's audio and transcript assumptions. Translation and transcription rates.

For 3.8 Flash TTS, Batch/Flex is $0.25 text input / $4.50 audio output per 1M tokens through December 31, then $0.50 / $9.00 from January 1. For 3.8 Flash-Lite TTS it is $0.25 / $3.00, then $0.50 / $6.00. For 3.1 Flash TTS Preview, Batch is $0.50 / $10.00. The current pricing page does not list context-caching charges for these TTS models. Google's TTS prices.

Free tier, grounding, and billing limits

The Standard free tier covers the Flash, Live and TTS input/output rows above, subject to model-specific limits; Argon and 3.1 Pro Preview have no free tier. Free-tier content is marked as usable to improve Google's products, while paid-tier content is marked as not used that way. Batch/Flex has no free tier for 3.8/3.6 Flash. Google's 3.5 Flash-Lite Batch/Flex rows separately mark free input/output, with no free caching: do not extrapolate that exception to other models. Free and paid terms by model.

Search grounding, which checks an answer against search results, is another overage surface. Google's paid rows include 5,000 free search requests per month shared across all Gemini 3 and newer models, then list $14 per 1,000 requests. Its footnote says each individual search query is charged; one customer request can trigger several queries. Track the underlying searches, not just user messages. Google's grounding terms.

Actual request and token limits appear in Google AI Studio. New paid setup requires linking a billing account and prepaying at least $5. Monthly billing-account caps are $250 on Tier 1, $2,000 on Tier 2, and $20,000 to $100,000 on Tier 3, across linked projects; you can set separate project caps. Tier 2 qualification is $100 paid plus 3 days from the first successful payment, and Tier 3 is $1,000 paid plus 30 days. Google's billing guide.

Spend-rate limits, where they apply, are $10 per rolling 10 minutes on Tier 1, $50 on Tier 2, and $200 on Tier 3. Batch allows 100 concurrent requests, a 2 GB input file and 20 GB of file storage. Google's rate limits.

Gemini pricing for Google Workspace

Business buyers have a third price ladder. On a one-year commitment, Google Workspace Business Starter costs $7 per user per month, Business Standard costs $14, and Business Plus costs $22. Starter includes 30 GB of pooled storage per user, Standard 2 TB, and Plus 5 TB. Google advertises a 14-day trial. These are regular annual-commitment prices, rather than temporary introductory offers. Google's Workspace pricing.

Gemini is included rather than sold as the old standalone Workspace add-on. Starter includes Gemini in Gmail plus basic Gemini app access. Standard and Plus expand Gemini across Gmail, Docs, Meet, and other Workspace apps.

The commitment is the hidden cost. Business Standard is $16.80 per user on flexible monthly billing versus $14 with a one-year commitment. Ten seats therefore cost $28 less per month, or $336 less per year, on annual commitment. Five seats cost $70 per month on annual commitment versus $84 flexible. The saving is real, but so is the lock. A growing company with uncertain headcount may reasonably pay $2.80 more per seat for flexibility. Google's billing-plan comparison.

Do not compare Workspace seats with consumer family sharing as if they were interchangeable. Workspace pricing buys managed business accounts and organization features. Family sharing distributes a consumer bundle. The Developer API remains separate from both.

Hidden costs, refunds, annual billing, and the student offer

Gemini's largest hidden cost is not a surprise surcharge. It is buying the wrong meter. The consumer app, Flow credits, Developer API, Google Cloud, and Workspace seats can all produce separate charges. Paying $19.99 for Pro does not automatically create a prepaid API balance, and paying Workspace seat fees does not erase Flow or API usage. Eligible Google Cloud credits can offset API charges, but with Prepay they apply only while you maintain an active prepaid API balance. Google's credit rules.

Credit expiry is the next trap. Free daily Flow credits and paid monthly Flow credits do not roll over. Pro and Ultra can buy AI credits in eligible accounts and regions; Plus eligibility is disputed across the Google help pages linked above. Public support pages show example increments rather than dollar prices, and purchased credits may carry expiry terms. An allowance has zero economic value if it expires unused.

Google's current plan page displays AI Pro at $19.99 monthly. Its Pro membership guide confirms monthly and annual subscriptions, but the current public pages do not quote an annual dollar price. Compare your checkout quote with twelve monthly payments totaling $239.88. Ultra's public quotes are monthly. Google's plan page. Google One's terms promise at least 30 days' notice before a subscription price increase.

Refund terms are stricter than many buyers expect. Most Google storage-plan purchases are non-refundable. Cancelling normally leaves access active through the paid term. Purchases made through Google Play or Apple follow that seller's refund policy, so the place of purchase controls the route; regional consumer rights can create exceptions. Google's refund policy.

Eligible US higher-education students currently have a 12-month AI Pro trial at no cost. Google requires student verification through SheerID with a valid school email, a personal Google Account, and a qualifying payment method. YouTube Premium and Premium Lite are not included during the trial. Check eligibility and the renewal terms before signing up. Google's current student offer.

The dated API discount is another reason to separate subscription pricing from usage pricing. October's Flash token bill can be half January's for the same workload, while the consumer plan is priced on its own schedule. Use the renewal and usage rates that apply when you will actually do the work.

Gemini versus ChatGPT and Claude on normalized price

Gemini AI Pro costs $19.99 per month, includes Argon in the app, a 1,000,000-token app context window and 5 TB of storage, and can share eligible AI benefits with up to five other people. Storage and credits are shared rather than an independent pool for each person. Ultra costs $99.99 for 5x Pro usage or $199.99 for 20x. Google's plan page.

For a comparison with another consumer subscription, count only benefits that replace a bill or a workflow constraint you already have. Unused storage and expiring credits do not reduce your effective cost. A developer should compare token input, output, caching and grounding for the same workload separately from the app fee.

Gemini's $4.99 entry tier, Google storage, long context and family sharing can make it a good fit. Price does not settle model fit. The Gemini versus ChatGPT comparison and Gemini versus Claude comparison are the better next step when capability, writing style, or workflow matters more than the bill.

Who should pick which Gemini plan

A casual user should stay on Free. The plan already covers normal chat, light research, images, and a small daily Flow allowance. A solo operator who feels those limits weekly but does not work with enormous source sets should buy Plus at $4.99.

A research-heavy founder, senior operator, or analyst should choose Pro only if Argon access, the 1,000,000-token context, 1,000 Flow credits, 5 TB, Google app access, or family sharing replaces a real workaround or bill. A creator should model monthly Flow output before buying Ultra. The move makes sense when hundreds or thousands of videos have a commercial destination, not when Ultra is merely more comfortable.

A software builder should start with the free Developer API tier for prototyping, benchmark Flash-Lite and Flash against the task, and activate paid API billing when rate limits, data terms, or production reliability require it. An organization already managing staff through Google Workspace should compare Workspace seats first, then add consumer or API spend only for needs the business plan does not cover.

The decision flips at a measured constraint: a context size you can name, a Flow volume you can forecast, a storage bill the bundle replaces, or an API workload with a unit cost. If you cannot name the constraint, stay one tier lower. Before your next budgeting cycle, count a representative workload's input and output tokens and price it at Argon's December rate or Flash's January rate; compare the app allowance only if the work happens in the app.

Gemini pricing FAQ

Is Gemini totally free?

Gemini's core consumer app has a $0 plan, and Flash variants have a limited free Developer API tier. Model access, context, prompt capacity, Flow credits, and rate limits still apply, but a paid charge does not begin unless you deliberately subscribe or activate paid API billing.

Is Google Gemini worth paying for?

AI Plus is worth $4.99 when Free limits interrupt real work often enough to matter. AI Pro is worth $19.99 when Argon in the app, the 1,000,000-token context, recurring need for higher Flow capacity, 5 TB of storage, Google app access, or family sharing replaces another cost; otherwise stay lower.

What is the difference between Gemini Pro and Ultra?

AI Pro costs $19.99, supplies 1,000 Flow credits, and has a 1,000,000-token context. Both Pro and Ultra include Argon in the app. Ultra starts at $99.99 with 5x Pro app limits and 10,000 Flow credits, then reaches $199.99 with 20x Pro app limits and 25,000 credits. Ultra is a throughput and production bundle, not simply a smarter chat plan.

Can Gemini be billed yearly?

Yes. Google's Pro membership guide confirms an annual subscription option. Use your current annual checkout quote and compare it with $239.88 for twelve payments at $19.99. The public Ultra prices are quoted monthly.

Is Gemini free for students?

Eligible US higher-education students can currently claim a 12-month AI Pro trial at no cost after verifying student status and meeting Google's account and payment requirements. YouTube Premium and Premium Lite are excluded during the trial. Other students can use the standard Free plan.

Can I get a Gemini refund?

Most Google storage-plan purchases are non-refundable. Cancellation generally preserves access through the paid term, while purchases through Apple or Google Play follow that seller's refund policy.

Is Gemini API pricing included in Gemini Pro?

No. AI Pro includes consumer and certain AI Studio benefits, but the Gemini Developer API remains a separate bill for metered calls, caching, grounding and media. An app allowance is not unlimited API usage.

Did Gemini pricing change in 2026?

Yes. Argon now has Standard API rates of $2.00 input and $10.00 output per 1 million tokens through November 30, 2026, becoming $4.00 and $20.00 on December 1. Gemini 3.8 Flash and 3.6 Flash remain $0.75 input and $3.75 output through December 31, 2026, becoming $1.50 and $7.50 on January 1, 2027. The app subscription ladder is a separate price schedule.

Want a faster way to match each AI subscription to a recurring business outcome? Get the AI Tools Map for Business Owners.