Gemini: reviews and analysis

TLDR: Gemini is Google’s general-purpose assistant, and its edge has less to do with the model than with where it lives: inside Gmail, Drive, Calendar and Android itself. The free tier is good enough to work with, and paid plans start from $4.99 a month, under what its direct rivals charge. What its own users complain about most are the usage limits, which are metered by compute and come with no published number. If your working life already sits inside Google, this is the assistant that asks the least of you.

What Gemini is and how it works

Gemini is Google’s conversational assistant and the brand the company has folded all of its consumer AI into. It answers questions, drafts, summarizes long documents, translates and generates images, video and music. On top of that sit the named features: Deep Research, which goes out and returns a sourced report, Gemini Live for talking to it out loud, Canvas for working on a document side by side, and Gems, assistants you configure once with your own instructions and reuse.

What separates it from ChatGPT or Claude isn’t that feature list, which by now looks much the same across products. It’s distribution. Gemini isn’t a site you visit, it’s a layer sitting inside the tools you already work in: it summarizes a Gmail thread without leaving Gmail, drafts in Docs, pulls figures out of a Drive file, cross-references your calendar with your inbox. Through 2026 it’s also taking over from Google Assistant on Android, so the button you’ve been pressing for years on your phone becomes this one.

The free version with any Google account isn’t a shop window: it includes image generation, Deep Research, voice conversation and variable access to the Pro model. Plenty of people use it daily having never paid, and that shapes how its plans should be judged.

Worth separating from its command-line sibling: for actual programming Google ships a different product, Gemini CLI, a terminal agent that works on your project. The one reviewed here is the generalist, the browser tab and the phone app.

What it’s like to use day to day

The first thing you notice is what you skip: there’s nothing to upload. Asking it to find the attachment someone sent you in March, summarize a client’s last three emails or draft something from the numbers in a spreadsheet works because the material is already in the same house. That saved copy-and-paste is why many people end up using it even when they prefer another assistant’s answers.

On mobile the fit is sharper still. Taking over the system assistant slot turns a voice query while driving or cooking into the normal route rather than an experiment. Native apps cover web, Android, iOS and macOS, with a high App Store rating behind them.

There are three points of friction worth knowing before you pay. The first is the limits. Google stopped counting messages and now meters compute, which depends on how complex your prompt is, which model gets used and how long the conversation runs. The official documentation explains the mechanism but publishes no absolute figure, only per-plan multipliers, so you know your plan gives you twice or four times the free allowance without knowing four times what. The May 2026 cut left more than one subscriber with less than they had signed up for.

The second is sheer presence. The assistant turns up on every surface of Gmail, Docs and Drive, and a good share of users resent having it permanently in front of them. You can tone it down in settings, though the controls aren’t where you’d look for them. The third is privacy, and it deserves its own section because it affects both free and paid personal accounts.

What happens to your conversations

With Keep Activity enabled, the Gemini Privacy Hub says Google stores what you type and upload and may use it, with human review in some cases, to improve its services and train AI models. You can turn that setting off or open a Temporary Chat. New conversations then stay out of model training, although they are retained for up to 72 hours to provide and protect the service.

There is an easy-to-miss exception: if you submit feedback, Google may review and use the associated conversation even with activity disabled. The free chat helps improve the product, but that is not an unavoidable obligation for the user. Client documents belong in a work account with its own terms rather than a personal workspace.

Pricing and plans

The model is freemium, with a permanent free plan and subscriptions from $4.99 a month. That entry plan, Google AI Plus, got cheaper in June 2026 and doubled its included storage, which put it below the standard rate charged by the competition. The next step up, Google AI Pro, costs $19.99 a month and goes head to head with the ChatGPT and Claude subscriptions. Above that sits a heavy-use plan priced like a professional tool.

The arithmetic here differs from other assistants, because you aren’t paying for the AI alone. Every plan bundles Google One storage, so if you were already paying for space for photos and mail, that part comes off the real cost, and for a family sharing one plan the subscription justifies itself. What you can’t work out in advance is how much use each tier buys you, because compute-based limits come with no numbers, and that leap of faith is one rivals don’t ask for.

Who it’s for (and who it isn’t)

Gemini is for people who already live in Google. If your mail, your documents, your calendar and your phone are Google’s, this is the assistant that makes you do the least work to hand it context, and two decimal points in a benchmark won’t make up for that. It’s also the best door in for anyone who wants to try generative AI properly without putting a card down.

It isn’t the first pick if your work is writing or picking apart long text with surgical precision, where Claude still sets the bar. Nor if you value custom assistants and years of habit in the same interface, which is what keeps people on ChatGPT. And if your priority is that no conversation of yours feeds anyone, no big-platform assistant will fully settle that, this one included.

Alternatives to Gemini

The two serious rivals are ChatGPT and Claude, and choosing between the three depends more on where your work already lives than on which model scores best this month. If you’d rather not choose, YesChat puts several top models under a single fee. And if summaries are exactly what you don’t want and clean links are, Kagi answers a different intent. The detail is in Gemini alternatives, and the full category in our guide to the ai assistants.

Frequently Asked Questions

Is Gemini free?

Yes, there’s a permanent free plan with any Google account, and it isn’t token: images, Deep Research, voice and variable access to the Pro model are all included. Paid plans raise the usage limits and add storage and video.

How many messages can I send a day?

There’s no published figure. Google meters compute based on prompt complexity, the model used and the length of the chat, and only publishes how much each plan multiplies the free allowance. It’s the most repeated complaint among paying users.

What changes compared to Google Assistant?

Gemini is taking its place on Android through 2026. It handles long, ambiguous requests and holds a conversation, where the classic assistant did better with short, specific commands.

Can I use it with company documents?

Technically yes, and the Drive integration makes it easy. Check your account’s data policy first. On a personal account you can turn off Keep Activity, but confidential material belongs in a workspace covered by business terms.