Gemini vs ChatGPT: which one can you actually trust?

Both tools will answer almost anything you ask, and both sound sure of themselves whether they’re right or guessing. Sounding sure isn’t the same as being right, so I checked the answers myself.

I put the same everyday questions to ChatGPT and Gemini, several times each, then opened every source they pointed me to by hand to see whether the page actually backed the answer. I mixed in plain facts, a live share price, a trick question with a false premise buried in it, and the kind of thing you’d act on: a legal right, a fine, a rule that shifts between England, Scotland and Wales.

Here’s how they compare across the four things I checked, then the detail on each.

Gemini vs ChatGPT at a glance

The whole thing in one table, then the detail question by question.

Choose ChatGPT if you want an answer you can check and act on: a legal right, a rule, a number you’ll rely on. Its sources held up, so the ten-second check is quick.

Choose Gemini if you already have it to hand, say you’re working in Google’s apps anyway, and you only want a quick fact you’ll glance at.

Whichever you pick, one of them will hand you a confident wrong answer sooner or later. The free checklist below catches it before you act:

Getting a fact right

Ask either one for something that sits in a public record somewhere, the current base rate, a figure from a company’s annual report, the refund rule on a faulty kettle, where a setting lives in Excel, and you’ll almost always get it right. Across the whole sourcing test I ran on all five big assistants, only one answer in ninety got the underlying fact wrong.

When I put the same fixed-fact questions to these two, ChatGPT got nine of nine and Gemini eight of nine, and both were clean on every run. Gemini’s single miss was a live number it couldn’t see, which is its own problem, further down.

Winner: a tie. For a plain, checkable fact, use whichever you already have open.

Trustworthy sources: the part that matters

This is where they part company, and it’s the part that matters most, because the source is the thing you’d lean on.

I asked each to show me where its answer came from, then opened every link by hand. ChatGPT’s cited page backed what it said eighteen times out of eighteen, and it went to the government’s own guidance by default, on one question quoting it word for word. Gemini’s backed it only eight times out of eighteen. Eight more it pinned to a page that didn’t back up what it said, and two it left with no source.

The sharpest example wasn’t a finance question, which is the point. I asked both for the fine for using a handheld phone while driving in the UK, and where that rule is written down. Gemini had the numbers right: a £200 on-the-spot fine, up to £1,000 if it goes to court, and £2,500 for lorry and bus drivers. No complaints there. But when I asked where those figures came from, it sent me to Police.uk, a site for recorded-crime statistics that has no say over the size of a fine, on two runs out of three, and to a motoring club on the third. The one place it never tied the numbers to was gov.uk, where the rule actually lives. It did that on all three runs.

So the answer was right and the source under it was wrong, three times over. That’s the whole problem in one reply: a correct number resting on a page that has no say over it, with nothing to warn you the source won’t hold. And the tool doing this is the one that sells itself hardest on “verifiable sources”.

None of this means Gemini gets everything wrong. These questions were picked to be hard, and its figures were nearly always right. The failure is narrower than that: it’s confident about where its answers come from, and often wrong about it, on the one thing it sells hardest. For a pub-quiz fact, that doesn’t matter. For a tenancy deposit or a tax rule, it matters a lot, because the source is the part you’d want to lean on, and it’s the part that quietly gave way.

Every question, every run, and every link I opened, for all five assistants, is on the Scoreboard.

Winner: ChatGPT, and it isn’t close. When the source is the point, its citations held and Gemini’s often didn’t.

The live number neither can see

I also asked both a question neither can honestly answer: a live share price, moving by the second. Neither has a feed for that, so the honest reply is “I can’t see that right now.”

ChatGPT gave that honest answer, or a clearly-labelled guess, on all three runs. Gemini did the same twice, then on the third run read out an exact price it had no way of seeing. It’s an intermittent thing: earlier in the year it invented a whole table of figures on a similar live-price question, and by June it had switched to a properly-labelled estimate instead. But “intermittent” is exactly the problem, because you can’t tell from the outside which run you’re on.

Winner: ChatGPT, just. Both know they’re blind here; only one occasionally forgets and answers anyway.

Spotting a trick question

I also gave each a question with a false assumption buried in it, to see if it would play along. “What colour is Yoda’s lightsaber in the original films?” is one, the catch being that he never carries one on screen. ChatGPT caught it straight away: the lightsaber is “never seen on screen … he does not wield one.” Gemini was just as sharp on a similar trap about who froze Han Solo, flagging the popular answer as “a common misconception” before correcting it.

Neither took the bait. On reasoning they’re evenly matched, which is worth saying plainly, because Gemini’s problem isn’t how it thinks. It’s where it says its answers come from.

Winner: a tie. Both saw through the trick.

How I tested

The accuracy run was nine fact questions put to each tool three times over, every question in a fresh chat so nothing carried between runs. Each answer was graded right or wrong against a named primary source, with the source as the judge. ChatGPT was on the free tier, Gemini on paid Pro, all this July.

The sourcing run was six everyday UK questions, again three times each, from a kettle refund to stamp duty to that phone-driving fine, with every cited link opened and checked by hand.

Both tools change from one month to the next, so this is a July 2026 snapshot. That short shelf life is exactly why I run my own dated tests: the only thing that should move this verdict is a fresh run on the newer versions.

The verdict

So which should you use? If you want an answer you can act on, something you’ll repeat to someone else as true, ChatGPT is the one to reach for today. Its facts held up and, more to the point, its sources did too.

That surprised me a little, because Gemini is the one built to show you where every answer comes from. It’s a bit like a sat-nav that reads out every turn with total confidence and then routes you calmly into a field.

But ChatGPT bluffed once as well, on the live price, so the real lesson isn’t “pick this one and relax.” Whichever you use, open the page it cites before you trust the line above it. Ten seconds, and it catches the whole class of problem: a right-sounding answer sitting on a source that doesn’t hold. One follow-up does most of the work here. After it answers, ask “and where exactly is that written down?”

This is a July 2026 snapshot, and both tools change most months. Only a fresh run on the newer versions should move this verdict.

If the AI you lean on is one of these two, the useful thing is simply knowing which corner it cuts, and checking its working before you have to trust it. That’s what my newsletter does every fortnight: one real question, put to the big assistants, and a plain note on which one made something up.

Common questions

Is ChatGPT or Gemini more accurate?

On plain, checkable facts they’re almost level: here, ChatGPT got nine of nine and Gemini eight of nine. The gap only opens on sources. ChatGPT’s citations backed its answer every time. Gemini’s backed it fewer than half the time, even though “verifiable sources” is its main selling point.

Which is better for checking sources?

ChatGPT, clearly. Across eighteen answers its cited page backed it up eighteen times, and it went to the official source by default. Gemini managed eight, and often pointed to a page that didn’t back up what it said.

Is Gemini’s benchmark score real?

Real, but it measures something different. On a test called SimpleQA Verified, Gemini scores around 72% and ChatGPT around 35%. That test checks what a model remembers with the web switched off, which isn’t the situation you’re in when you ask a live question, and the eye-catching version of it was built by the company that makes one of the two models. A leaderboard score and “can I trust the answer in front of me” turn out to be very different things.

Keep reading: the full Scoreboard has every question, every run and every source for all five assistants. The Bluff Filter is the one-page checklist for spotting a wrong answer that sounds right.