benchr
Here's the thing nobody planned: the three big consumer assistants now cost almost exactly the same. ChatGPT Plus is $20 a month. Claude Pro is $20 a month, or about $17 if you pay for the year. Google AI Pro is $19.99. So the old tiebreaker, price, is gone. You're not choosing a cheaper plan. You're choosing a default model and the app it lives in.
And the defaults are where this gets interesting, because the three companies point you at very different models the moment you open the box.
| Assistant | Paid plan | Default model | What the free tier gives you |
|---|---|---|---|
| ChatGPT | $20/mo (Plus) | GPT-5.5 Instant | GPT-5.5 with a roughly 10-message-per-5-hours cap, then a smaller fallback model |
| Claude | $20/mo, $17/mo annual | Claude Sonnet 4.6 | Sonnet 4.6 with web search, memory, and voice; usage capped per session |
| Gemini | $19.99/mo (Google AI Pro) | Gemini 3.5 Flash | Flash chat plus a daily allotment of Gemini 3.1 Pro, image generation, and a few deep-research reports |
Notice the pattern. ChatGPT hides the model choice behind one auto-switching system, so you never think about it. Claude points you at its fast, balanced Sonnet tier and keeps the heavyweight Opus 4.8 for paid plans behind a selector. Gemini runs a quick Flash model for most turns and hands you a daily ration of its big Gemini 3.1 Pro for the hard stuff. Three philosophies, one price.
To make this concrete, four everyday tasks, scored. Not coding benchmarks. The actual things you'll do this week.
Task one: a quick answer you can check
You want a fact, a definition, or a "what changed this week" answer, and you'd like to see where it came from. All three now search the web, but they present sources differently. Gemini leans hardest on links, because that's Google's whole instinct, and it tends to surface fresh results fast. ChatGPT gives you a clean answer with a Sources panel you can expand. Claude searches too, but it's the most likely of the three to tell you when it isn't sure.
If checking sources is the point, a dedicated answer engine still beats all three for this one job. We pull that apart in the AI search engines comparison. For a general assistant you'll keep open all day, Gemini edges this task on speed and link density.
Winner: Gemini, on link-first answers.
Task two: drafting an email or message
Write the awkward reply. Decline the meeting without burning the bridge. Turn three bullet points into a note that sounds like you. This is where Claude pulls ahead, and it isn't close. Sonnet 4.6 lands a human, plain tone on the first try, where ChatGPT often reaches for the corporate register ("I hope this email finds you well") until you tell it not to. Gemini sits in the middle, competent and a little generic.
Anthropic's own framing for Sonnet 4.6 is instruction-following and consistency, and you feel it here: tell it "warmer, shorter, no exclamation points," and it holds all three. If most of your day is messages and short-form writing, this task alone can decide the whole thing. The Sonnet 4.6 review goes deeper on where that tone discipline comes from.
Winner: Claude, by a clear margin.
Task three: explain something so it sticks
"Explain how a mortgage refinance actually works." "Walk me through this lab result." Teaching is a breadth game, and ChatGPT's GPT-5.5 is the strongest generalist of the three. It picks a sensible level, uses an analogy that fits, and stops before it drowns you. Gemini is close and sometimes better when the topic touches Google's index of fresh information. Claude explains carefully and flags what it's unsure of, which is reassuring for anything where a confident wrong answer would cost you.
For pure "make me understand this" sessions, ChatGPT is the one most people will find easiest to learn from. The GPT-5 review covers how that generalist polish holds up under harder, more niche questions.
Winner: ChatGPT, on breadth and pacing.
Task four: images and a casual creative spin
Make a birthday graphic. Mock up a poster. Riff on ten name ideas. All three generate images now, and Gemini has the most generous free-tier image access plus tight links into Google's creative tools. ChatGPT's image output is strong and the most "just works" for a quick one-off. Claude is the odd one out here: it's built for text, and image generation isn't its game, though it's the best of the three at then writing the caption.
For casual visual creation bundled into your assistant, Gemini gives you the most for the money. If your creative work is specifically video, that's a separate decision we cover in the best AI for video roundup.
Winner: Gemini, on free-tier creative range.
Quick checkable answers
Gemini Link-first, fastDrafting and messages
Claude Human tone, first tryExplaining to learn
ChatGPT Best generalist pacingImages and creative
Gemini Most free-tier rangeTwo-one-one on the scoreboard, with Gemini taking two of the lighter tasks and Claude and ChatGPT splitting the two that matter most to heavy users. That's a useful summary, but the scoreboard isn't the recommendation. The recommendation depends on which of these four you do every single day. So here are the three verdicts, plainly.
You want one app and never want to think about which model you're using. GPT-5.5 Instant is the strongest generalist, the auto-switcher hides every decision, and it's the safest thing to put in front of someone who just wants answers. It's the default recommendation for a reason.
Your day is writing, editing, and replies you want to sound like a person. Sonnet 4.6 holds tone and instructions better than the other two, and Pro unlocks the heavyweight Opus 4.8 when you need it. If you write for a living, stop reading and subscribe here.
You already live in Gmail, Docs, and Search, and you like fresh, link-backed answers plus generous image access. Google AI Pro is the value play of the three, and the daily slice of Gemini 3.1 Pro covers the harder questions when Flash runs out of room.
One more piece of advice that holds across all three: don't pay for two. Subscribe to the one that fits your main job, keep the other two on their free tiers for the occasional second opinion, and check back in a few months, because the defaults change fast. If your real question is which underlying frontier models are pulling ahead, the head-to-head in Opus 4.8 vs GPT-5.5 and the per-model Gemini evaluation go a level deeper than any consumer app will show you.