Gemini Pricing 2026: Free, AI Plus, Pro & API Costs Compared – DIY AI

Published on September 15, 2026 by Steven Jones

Google Gemini has a free tier, Google AI Plus, Google AI Pro, two AI Ultra levels and separate pay-per-token API billing. On current US-dollar plan pages, AI Plus is $9.99 per month for the 2 TB configuration, AI Pro is $19.99 per month, and AI Ultra costs $100 or $200 per month depending on the usage tier. The API is billed separately.

The useful way to compare these plans is not simply “free model versus paid model”. Free Gemini users can still access Google’s current Gemini model family. Paying mainly buys more compute allowance, a larger context window and extra Google services. If you are only pricing Gemini because you need creative generation rather than a general AI assistant, see DIY AI Studio plans – from $9.99/mo, 100 credits. DIY AI Studio covers image, video, music and voice generation under one credit pool, but it does not replace Gemini’s text chat, research or Google Workspace features.

Prices and limits below were checked on 15 September 2026. Google varies storage configurations, promotions and local pricing, and Gemini usage is now measured through compute-based limits rather than a fixed number of prompts. That makes the headline monthly fee only part of the buying decision.

Gemini pricing at a glance: paid plans mostly buy usage headroom

Plan	Current price	Main allowance difference	Best fit
Gemini Free	$0	Standard compute limits, 32K context	Occasional chat, research and light everyday use
Google AI Plus	$9.99/month for the current 2 TB US-dollar plan	About 2x standard usage, 128K context	Regular users who want more headroom without paying for Pro
Google AI Pro	$19.99/month	About 4x standard usage, 1M context	Frequent Gemini users, long documents and Google-first workflows
Google AI Ultra	$100 or $200/month	Roughly 5x or 20x AI Pro usage, 1M context	Heavy users who can put a real value on avoiding limits
Gemini Enterprise	Custom	Organisation controls, support and enterprise deployment options	Businesses that need managed deployment rather than a personal subscription
Gemini API	Usage based	Separate token, caching and tool charges	Developers building Gemini into products or automations

The biggest trap in the pricing table is assuming that AI Plus, Pro and Ultra correspond neatly to progressively “better” chat models. They do not. Google currently gives free users access to Gemini Flash-Lite, Flash and Pro, but with smaller allowances. The practical upgrades are capacity and context: Free gets a 32K context window, AI Plus gets 128K, and AI Pro and Ultra reach 1 million tokens.

Google also changed how Gemini limits work. Usage is compute-based, so a long conversation using expensive features can consume allowance faster than a short text prompt. Limits refresh periodically, but Google can change them, and the remaining value of a subscription is harder to predict than it was under a simple prompt-count model.

A recurring pattern in subscriber discussions is that AI Pro is easier to justify when the bundled storage and Google services already have value. People buying purely for AI tend to be much more sensitive to changing usage caps. Ultra therefore makes most sense as a capacity purchase: pay for it because hitting limits costs you time or money, not because the price alone implies every answer will be dramatically better.

---

---

Gemini API pricing by model: output volume can dominate the bill

A Gemini consumer subscription does not give you unlimited Gemini API usage. Developer billing is a separate cost model, so someone paying for AI Pro can still incur API charges when an app, script or automation sends requests through the Gemini Developer API.

API model	Input price	Output price	Important condition
Gemini 3.8 Flash Standard	$0.75 per 1M tokens	$3.75 per 1M tokens	Current promotional rate runs through 31 December 2026, then rises to $1.50 input and $7.50 output
Gemini 3.1 Pro Preview, prompts up to 200K	$2.00 per 1M tokens	$12.00 per 1M tokens	No free API tier for this model
Gemini 3.1 Pro Preview, prompts over 200K	$4.00 per 1M tokens	$18.00 per 1M tokens	Long prompts move into a more expensive pricing band

Google’s Gemini Developer API pricing also includes separate rates for context caching, grounding and other tools. Batch processing can cut token charges by 50% for work that does not need an immediate response.

The cost shape matters more than the headline input rate. At the current September 2026 Flash promotional rate, 10 million input tokens plus 2 million output tokens cost about $15. The same token volumes on Gemini 3.1 Pro Preview, assuming prompts remain at or below 200K tokens, cost about $44. Pro becomes even more expensive when individual prompts cross the 200K threshold.

That is why output discipline is a real API cost control. On Flash, output tokens cost five times as much as input at the current promotional rate. On Pro’s lower context band, they cost six times as much. For classification, extraction or background processing, asking for a concise structured response can matter more to the bill than shaving a modest amount off the prompt.

If you are comparing providers rather than only Gemini tiers, our AI model cost comparison is the more useful place to compare inference pricing and model trade-offs.

Is Gemini free enough? Use context size and usage pressure as the decision test

Gemini Free is more capable than the usual “free trial” framing suggests. Because it still includes access to Google’s current model family, the right upgrade trigger is usually not “I want Gemini Pro”. It is “the free limits are now interrupting work I actually care about”.

• Stay on Free if you use Gemini intermittently, rarely hit limits, and don’t need to work with very large files or long-running contexts.
• Choose AI Plus if you want a lower-cost buffer above Free and 128K context is enough. It is the sensible middle tier for regular personal use.
• Choose AI Pro if 1M context, heavier usage and the broader Google bundle are useful every week. At $19.99, this tier should be compared directly with other $20 AI subscriptions.
• Choose Ultra only if you can measure quota pressure. If you cannot point to work you currently abandon, delay or move elsewhere because Pro runs out, the extra $80 to $180 per month is difficult to defend.

Another hidden difference: Pro and Ultra users can buy additional AI credits for some Google AI products, while AI Plus is a more fixed-capacity plan. For unpredictable heavy workloads, that flexibility can be more valuable than a slightly cheaper base subscription.

The common mistake is upgrading for model prestige. A better test is to keep Free until you can name the constraint: context length, usage allowance, a Google Workspace feature or a bundled service. Then pay for the lowest tier that removes that specific constraint.

Gemini vs ChatGPT Plus vs Claude Pro: the $20 tier is the real comparison

Gemini AI Plus looks cheaper than ChatGPT Plus or Claude Pro because it is. It is also a lower-priced tier. The cleaner comparison is Gemini AI Pro at $19.99 per month against ChatGPT Plus at $20 per month and Claude Pro at $20 per month.

Subscription	Monthly price	Best reason to choose it	Main pricing caveat
Google AI Pro	$19.99	1M context plus a Google storage and app bundle	Gemini API usage is still billed separately
ChatGPT Plus	$20	Users whose work already centres on ChatGPT’s tools and custom workflows	OpenAI API usage is separate from the Plus subscription
Claude Pro	$20, or roughly $17/month on the annual plan	Users who prefer Claude’s writing, coding and document workflow	Anthropic API usage is also separate

At roughly $20, price shouldn’t drive this choice. Gemini Pro has the clearest financial case for someone already paying for Google storage or using Gmail, Docs and other Google services heavily. ChatGPT Plus makes more sense if your existing custom assistants and tool workflows live in ChatGPT. Claude Pro is easier to justify when Claude is already your preferred writing, coding or long-document environment.

The useful shortcut is to compare the subscription that fits your workflow, then treat API spend as a separate budget line. Mixing consumer subscription prices with developer token costs creates a misleading comparison because none of these $20 plans is simply an API allowance in disguise.

Your search, your sources

Make DIY AI a preferred source

See more of our reporting in Google Top Stories, AI Overviews and AI Mode.

Add DIY AI on Google

For creative generation, a shared credit pool can cost less than stacked subscriptions

Gemini is the better fit if you need a general-purpose assistant, research, long-context analysis or deep Google Workspace integration. DIY AI Studio is not a substitute for those jobs.

The comparison changes if most of your spend goes into media generation. DIY AI Studio Starter costs $9.99 per month for 100 credits shared across image, video, music and voice generation. That can be simpler than managing separate subscriptions for each creative category, especially if your usage shifts between formats month to month.

Video is a good example of why this matters. A general AI subscription may give you some generation allowance, but serious video workloads can consume quota quickly or push you towards a second paid product. Our Google Veo pricing breakdown shows how quickly model-specific generation costs can become a separate budget decision.

The sensible setup for many users is therefore not “Gemini or Studio”. It is Gemini Free or Pro for assistant work, then a dedicated creative budget only if image, video, music or voice generation is where the actual production volume sits. That avoids paying for a premium AI assistant tier just to chase more media allowance.

Which Gemini plan should you pay for?

Most people should start with Gemini Free and upgrade only after a concrete limit appears. AI Plus is the low-cost step up for regular use. AI Pro is the strongest general-value tier for heavy Google users because it combines 1M context, higher compute allowance and the wider Google bundle at roughly the same monthly price as ChatGPT Plus and Claude Pro.

Ultra is different. At $100 or $200 per month, it should solve a capacity problem you can already quantify. If Pro’s limits are not costing you productive time, Ultra is usually paying for theoretical headroom.

Developers should budget the API independently. Use Flash where latency and cost matter most, move to Pro only where the task benefits enough to justify the higher token rate, and use batch processing for non-urgent workloads. That is a more reliable way to control Gemini spend than choosing the most expensive subscription and assuming everything else is included.