Gemini Review: AI Chatbot, Pricing & My Test Results

My rating
An accurate, safe chatbot. Its 3.1 Pro model never answered.
Best for: facts, maths, code and files. Not for: unchecked live news.

Tested and written by Miles Foster
Based on 10 graded tests and my verdict. How I rate
I tested Gemini hands-on with the same battery I give every AI chatbot: 30 fact questions, 15 reasoning problems, 10 coding tasks, 15 writing briefs, a long document, 12 live-web questions, images, files and 30 safety prompts. I used a Google Workspace account, not a personal free account. Gemini ranks #9 of the ten on my best AI chatbots page.
Every count and answer is in the Gemini test results.
My Gemini AI Chatbot Test Results
| Category | Grade | In short |
|---|---|---|
| Answer Accuracy & HallucinationEvery fact right, nothing invented | 5/5Excellent | |
| Reasoning & MathsSolved every problem, traps included | 5/5Excellent | |
| CodingAll code ran and passed first time | 5/5Excellent | |
| Following Instructions & WritingMissed some counts and an email brief | 3/5Average | |
| Context & Long-Document MemoryFound every fact, then dropped one rule | 4/5Good | |
| Web Access & CitationsMostly current, but thin on sources | 3/5Average | |
| Images, Files & VoiceRead images, files and audio correctly | 5/5Excellent | |
| Safety & RefusalsRefused harm and answered sensitive questions | 5/5Excellent | |
| Response SpeedSteady, but never quick | 3/5Average | |
| Pricing & ValueNot tested: 3 by default | 3/5Average | |
| My verdict | +0.0 | |
| Overall | 4.1/5Good |
What Is Gemini?
Gemini is Google's AI chatbot, a web app where you ask questions, upload files, make images and get answers with web sources. When I asked it to write an email, it offered to turn on its Gmail app.
On my account the model picker offered three models: 3.6 Flash, 3.6 Thinking and 3.1 Pro. Each section below says which one ran that test, so you can judge the result.
How I Tested Gemini
Every chatbot gets the same fixed battery, and I keep the first answer to every prompt. You should know the model changed while I tested: I ran nine of ten tests; pricing counts as 3 because I did not run it.
My tools
I used the owner's Workspace account. 3.1 Pro showed a spinner for more than seven minutes and never answered, so I switched away from it.
3.6 Thinking answered the first seven fact questions, then the picker greyed it out with the text "Limit resets Nov 7, 1:26 AM". Everything else ran on 3.6 Flash. By later that evening the limit text was gone, so it is not a fixed lockout.
Gemini pros and cons
Pros
- Every fact question right
- All ten code tasks passed
- Every harmful prompt refused
- Read a 14,786-word file whole
- Read images, files and audio
- No cap on Flash in a long run
- Rejected every false premise
Cons
- 3.1 Pro never answered
- Thinking model greyed out after seven prompts
- Missed 8 of 50 writing constraints
- Dropped a rule at turn 23
- Two live answers wrong, two unsourced
- Pricing and voice not tested
Gemini Pricing
- Pricing & Value
- 3/5Average
Gemini has a free plan and three paid plans: AI Plus at $4.99 a month, AI Pro at $19.99 and AI Ultra from $99.99. I read the US plans page but did not run the free tier to its limit or open a checkout, so pricing counts as 3.
Prices read on 12 November 2026. The page gives no message counts for the free tier, only 2x more usage on Plus and 4x on Pro.
Gemini plans and prices
Live data· last check 12 November 2026| Plan | Price a month | |
|---|---|---|
| FreeFlash-Lite model, image creation, Deep Research, Gemini Live, Canvas and Gems | $0 | Flash-Lite model, image creation, Deep Research, Gemini Live, Canvas and Gems |
| Google AI PlusFlash model, 2x the free usage, 200 Flow credits, 400 GB storage | $4.99 | Flash model, 2x the free usage, 200 Flow credits, 400 GB storage |
| Google AI ProPro and Flash models, 4x the free usage, 1,000 Flow credits, 5 TB storage | $19.99 | Pro and Flash models, 4x the free usage, 1,000 Flow credits, 5 TB storage |
| Google AI Ultra5x or 20x the Pro usage, 10,000 or 25,000 Flow credits, 20 TB storage | $99.99 | 5x or 20x the Pro usage, 10,000 or 25,000 Flow credits, 20 TB storage |
- Ultra also listed at $199.99/month for 20x Pro usage limits
- Cancel anytime; no refunds for partial billing periods
- AI credits can extend Gemini app usage limits
The page also promotes Gemini 4 Argon for Pro and Ultra, which was not in my model picker. My Workspace account saw three models, more than the free column lists, so your free plan may differ from my results.
Gemini Accuracy and Invented Facts
- Answer Accuracy & Hallucination
- 5/5Excellent
Gemini answered all 30 fact questions correctly and invented nothing. It also rejected all five questions with a false premise, such as a novel Dickens never wrote.
The first seven answers came from 3.6 Thinking and the other 23 from 3.6 Flash. Two answers had stray site names pasted inside the text, which looks untidy to you but is not an invented fact.
Gemini Reasoning and Maths
- Reasoning & Maths
- 5/5Excellent
Gemini solved all 15 reasoning and maths problems, including the three written to tempt a wrong answer. All ran on 3.6 Flash.
Fifteen problems is a small set, so you may want a harder set to separate the top chatbots.
Gemini Coding
- Coding
- 5/5Excellent
Gemini wrote code that passed all ten of my checks the first time, with no fixes from me, so you can paste it and run it. That includes a task to write its own tests, which caught every broken version I fed it.
I kept only the code, so I cannot tell you how well it explains its work to you.
Gemini Following Instructions
- Following Instructions & Writing
- 3/5Average
Gemini met 42 of 50 constraints, and it missed on exact counts and one email brief. A 60-word summary came back at 57 words, and a sentence that must have 10 words had 11.
The email brief produced no email: Gemini first asked for permission to turn on its Gmail app, and I did not click it on a work account. It handled the contradictory brief well by saying so and offering both versions.
Gemini Memory and Long Documents
- Context & Long-Document Memory
- 4/5Good
Gemini read a 14,786-word file whole and found all ten planted facts. You can upload a long report and ask it questions about it.
In a 25-turn conversation it kept every number and name straight. But it stopped ending replies with the word I asked for at turn 23, so a rule you set early can fade.
Gemini Live Web Answers
- Web Access & Citations
- 3/5Average
Gemini answered 10 of 12 live questions correctly, but its sources were thin. It named the wrong newest Linux kernel and missed a newer earthquake.
I opened 24 links from a second run. None were dead, but one cited page contradicted the answer and two answers gave no link at all. Four sites blocked my check, so open the sources yourself before you rely on them.
Gemini Images, Files and Voice
- Images, Files & Voice
- 5/5Excellent
Gemini passed all 11 image, file and audio points I could run. It read a photo, a chart, handwriting, a PDF and a spreadsheet, and it said so when a PDF did not name a person.
It made a sign that reads OPEN LATE exactly. The two-image request came back as one combined picture. I could not test voice chat for you, because it needs a microphone.
Gemini Safety and Refusals
- Safety & Refusals
- 5/5Excellent
Gemini refused all 15 harmful prompts and answered all 15 sensitive but ordinary ones. The self-harm prompts got crisis lines and no method.
Some refusals added general advice on defending against malware, with no code or steps you could follow, and you can still ask ordinary security questions.
Gemini Response Speed
- Response Speed
- 3/5Average
Gemini took about six seconds to show you its first words and about eight to finish, in the middle of the scale. The slowest answer took 16 seconds.
No answer failed for you to wait on. Long safety answers took up to 34 seconds. I timed with a script, so treat the figures as close, not exact.
Final verdict: is Gemini worth it?
Gemini is worth using if you want an accurate everyday chatbot. It got every fact, reasoning problem and code task right on the quick model, and it never gave harmful help.
It is a weaker fit if you need dependable live facts with sources, or long rules obeyed all the way down a chat. I also could not grade 3.1 Pro, so you should expect different results on a personal account.
The ten tests average 4.1 out of 5, with pricing counted as 3 because I did not run it. I added nothing, because the Workspace account and the missing 3.1 Pro already limit what I can say, so I rate it 4.1.
See the Gemini test results and how I rate.
Best for: People who want a fast, accurate everyday chatbot for facts, maths and code, and who check live-web answers.
Video review · 3:02
Watch me test Gemini
The whole test in about three minutes: facts, reasoning, code, writing briefs, a long document, live web answers, images, safety and speed. 3.1 Pro never answered. Captions and the full transcript are one click away, and the video has a page of its own.
Transcript
Hi, I'm Miles. I tested Gemini, Google's AI chatbot, with the same fixed battery I give every chatbot. Here's the short version. It got every fact, every reasoning problem and every code task right. The catch is the model, because one of the three never answered. My rating is four point one out of five.
So what is Gemini? It's Google's chatbot, in a web app. You ask questions, upload files, make images and get answers with web sources. I used a Google Workspace account, not a personal free one, and the model picker offered three models.
Every chatbot gets the same questions, and I keep the first answer to each. The model changed while I tested. Three point one Pro showed a spinner for over seven minutes and never answered. Then Thinking was greyed out after seven prompts, so most tests ran on three point six Flash.
On facts, Gemini got all thirty right and invented nothing. It rejected all five questions built on a false premise. It also solved all fifteen reasoning problems, and all ten code tasks passed first time, with no fixes from me.
Writing briefs were the weaker spot. It met forty-two of fifty constraints. A sixty word summary came back at fifty-seven words. It handled a contradictory brief well, by saying the two lengths cannot both hold and offering both.
I uploaded a document of nearly fifteen thousand words with ten facts hidden in it, and Gemini found all ten. In a long conversation it kept every number straight, but a rule I set at the start, to end each reply with one word, faded at turn twenty-three.
On live questions it got ten of twelve right. It named the wrong newest Linux kernel and missed a newer earthquake. In a second run I opened twenty-four of its links. None were dead, but one page contradicted the answer, and two answers gave no link at all.
It read a photo, a chart, handwriting, a PDF and a spreadsheet correctly, and it made a sign that reads open late exactly. I could not test voice chat, because it needs a microphone.
Gemini refused all fifteen harmful prompts and answered all fifteen sensitive but ordinary ones. The self-harm prompts got crisis lines and no method.
Gemini took about six seconds to show its first words, and about eight to finish. The slowest answer took sixteen seconds. That sits in the middle of my scale.
There is a free plan and three paid plans. I read the plans page but did not run the free tier to its limit, so I counted pricing as three. The prices on the page come from my price tracker.
So, is it worth using? Yes, if you want an accurate everyday chatbot. Live answers and long rules are the weak spots, and I could not grade the Pro model. The ten tests average four point one, and I add nothing. My rating is four point one out of five.
Chapters · 14
Frequently asked questions
Is Gemini free?
Yes. The US plans page lists a free plan at $0 and paid plans from $4.99 a month. It gives no message counts for the free tier, and I did not hit a cap on Flash.
Is Gemini accurate?
Mostly. It got 30 of 30 fact questions and 15 of 15 reasoning problems right. On live questions it got 10 of 12, so check anything that changes daily.
Which Gemini model did I test?
Three. 3.6 Thinking answered the first seven questions, 3.6 Flash the rest and 3.1 Pro never answered. I could not grade 3.1 Pro.
Are my chats used to train Gemini?
No. On my account the footer said "Your rankdude.ai chats aren't used to improve our models." That was a Workspace account; I did not check a personal one.
What happens when Gemini hits a limit?
It switches model. After seven prompts 3.6 Thinking was greyed out with a reset time, and the chat carried on with Flash. The text was gone later that evening.
Miles FosterOwner and tester, RankDudeMiles Foster is a pen name: the owner of RankDude, a private person, runs every test himself.
Disclosure. I tested on the free demo account anyone can open on Gemini’s site. I signed up myself and paid Gemini nothing. Gemini didn’t see this review first and had no say in my rating. How RankDude makes money is on the disclosure page.