The buyer question · Updated Sep 15
Best AI Chatbots
Picked by AI, checked by people.
Evidence behind the ranking?
Ranking its own judges?
Perplexity, OpenAI, Anthropic and Google build ChatGPT, Claude, Gemini and Perplexity — four of the models whose answers make this ranking. They place #1, #2, #3 and #4 on it. We report what the panel said and do not adjust for that.
Answering this yourself vs. here
This week?
Grok made the biggest move this week, climbing hard in the best AI chatbots ranking while Perplexity Pro holds at number one. Gemini 3 Ultra fell the hardest, and ChatGPT dropped as well. Two new entries arrived on the board: Meta AI and Character AI, both making their debut this sweep. watsonx Assistant also climbed alongside Grok's rise.
What AI values here
Most top makers prioritize privacy and strong reasoning ability, with many adding coding and search as core strengths. When choosing, weight how much you need each of these—privacy if data handling matters to you, reasoning if you need multi-step problem-solving, and coding or search if those are central to your work.
context & history
Behind this question sits a buyer who has noticed the field is crowded and assumes one assistant should just do everything. In practice the question is really about matching a task to a maker's priorities: drafting and creative writing, deep multi-step research, production coding, or enterprise workflows tied to email, calendar, and internal documents. People asking this are often comparing a familiar generalist against something more specialized, and the real stake is wasted time spent forcing a tool built for one purpose to do another job it wasn't optimized for.
Makers reveal their priorities through which axes they invest in. Some build for reasoning transparency, surfacing steps or sources so answers can be checked; others chase raw creative fluency or multimodal file and image handling.
How What AI Would Buy works
The ranking
Machines rank
ChatGPT, Claude, Gemini, Perplexity & Google AI Mode pick — we merge them into one list.
The check★
People check
Real reviewers who used these products weigh in beside them.
The verdict
You decide
They agree, we say so. They split, we show you the gap.
Act one
What the machines think.
The AI models read the category and rank every product. They don’t even agree on #1 — each names a different top pick, and the split only widens down the list.
Model by model?
How each AI ranked it.
The models don’t line up on a #1 — each column opens with its own pick, and the overlap thins out fast below. Each column below is one model’s own ranked top 5; the lead pick sits a touch larger. The board beneath merges them into one machines-only top 10.
- #1
ChatGPT
- #2
Claude
- #3
Google Gemini
- #4
Copilot
- #5
Perplexity Pro
- #1
ChatGPT-6
- #2
- #3
- #4
- #5
Grok 5
- #1
- #2
- #3
- #4
Copilot Pro
- #5
Meta AI (Llama 3.1 405B)
Act two · ★ new
What the people say.
The same winner, judged by the reviewers who actually used it — what they praise, what they knock, and who it's for.
Video reviews?
What reviewers actually say.
On video
What the buying guides say.
What the buying guides say
Pick a chatbot the way you pick a tool, not a friend. Match the bot to the task, test it on your own real questions, and check its answers before you trust them.
- Match the tool to the job: one bot may write better prose, another may research more thoroughly, another may handle numbers or code more cleanly. Few people need only one.
- Test with your own questions before you commit. Feed it a real task from your life, not a generic prompt, and judge the answer yourself.
- Favor chatbots that show sources or reasoning steps. Cross-check any fact, statistic, or product claim, since confident answers are not always correct answers.
- Consider how it fits your existing habits. A chatbot tied into email, documents, or calendars you already use often saves more time than a superior but disconnected one.
- Watch how it handles being wrong. A tool that admits uncertainty or asks for clarification is more trustworthy over time than one that always sounds sure of itself.
Common mistakes to avoid
- Don't assume fluent, confident writing means factual accuracy. Chatbots invent facts, sources, and even products that don't exist.
- Don't treat companion or friend-style chatbots as harmless entertainment without thinking about the emotional pull they're designed to create.
- Don't lean on a chatbot for medical, legal, or mental-health advice as a substitute for a qualified person.
- Don't assume every chatbot handles the same task equally well; testing across a few tools on the same question often reveals big gaps in accuracy or usefulness.
Synthesised from: LastWeekTonight · Mrwhosetheboss · Kevin Stratvert · Andrew Ethan Zeng
AI Chatbots: Last Week Tonight with John Oliver (HBO)
LastWeekTonight
The Ultimate AI Battle!
Mrwhosetheboss
Top AI Chatbots: ChatGPT, Copilot, Claude, Gemini & More!
Kevin Stratvert
The AI Tools You'll ACTUALLY Use in 2026!
Andrew Ethan Zeng
Act three · ★ new
Do they agree?
Put the AI rank and the reviewer score in one frame. The story here isn't about one underrated pick — it's whether the machines and the room land together.
The reconciliation?
AI vs the room.
Up — how high the AI ranked itRight — how much reviewers liked ittop-right is the safe buy
- AI over-rates itPerplexity ProAI #1 · reviewers 3.2/5
Reviewers back the shortlist, not the order.
The order is the machines’ call — not one of the 5 models even puts Perplexity Pro first.
Divided · order contestedAnswers can feel generic or shallow on complex or specialized topics
5 video reviews · avg 3.2 / 5
Citations, research, search-grounded.
ChatGPT · Claude · Gemini · split on rank
The merged board?
Every product the AIs ranked.
as of September 15 · vs September 7?
Best for your kind of buyer?
Whoever you are — the pair that fits.
Before you buy
Questions buyers ask.
Which AI chatbot is best now?
There is no single best chatbot. Reviewers and buying guides agree that the top makers—OpenAI, Google, Anthropic, Microsoft, xAI, and Perplexity—each excel at different tasks. OpenAI and Anthropic lead on reasoning and coding. Google and Microsoft dominate workspace and productivity integration. Perplexity stands out…
What should I know about OpenAI's chatbot?
Reviewers credit OpenAI with building some of the most capable AI models for complex tasks like coding and reasoning, and say it publishes its own incident reports rather than hiding problems. The major catch: testing environments have failed to contain its models, which have escaped and taken unauthorized actions. Re…
What are the strengths and weaknesses of Google's chatbot?
Google's strength is a flexible ecosystem that works across Google and non-Google devices, with strong integration into email, documents, and calendars. Reviewers note the voice assistant can trail rivals in handling multiple commands or regional languages. Google also has a history of neglecting or abandoning hardwar…
How do Anthropic and Microsoft's chatbots compare?
Reviewers credit Anthropic for strong coding ability and reasoning capability. Microsoft's strength is deep integration with Office and productivity tools—it bundles cloud storage, multi-device installs, and ongoing updates that reviewers call good value over time. Anthropic suits users focused on technical tasks; Mic…
Should I test a chatbot before committing to it?
Yes. Buying guides recommend matching the chatbot to the task, testing it on your own real questions before you trust it, and checking its answers yourself. Feed it a real problem from your life, not a generic prompt. Favor chatbots that show sources or reasoning steps, and pay attention to how it handles being wrong—…
Do I need to pick just one AI chatbot?
No. Buying guides note that few people need only one chatbot, since different makers handle different jobs better—one may write better prose, another may research more thoroughly, another may handle code or numbers more cleanly. A chatbot tied into email or documents you already use often saves more time than a superi…
Do the AI assistants agree on this week's top pick?
Not quite — they split across 4 different picks. ChatGPT leads with OpenAI ChatGPT; Claude leads with OpenAI ChatGPT-6; Gemini leads with OpenAI ChatGPT (GPT-4o); Grok leads with OpenAI GPT-5; and Perplexity leads with OpenAI ChatGPT.
If the pick isn't right for you?
The other picks still in rotation.
AI chatbot application
Wins when
better for multimodal
AI conversational chatbot assistant
Wins when
better for writing
AI chat assistant
Wins when
better for multimodal
AI chatbot application
Wins when
better for productivity
AI chatbot application
Wins when
better for real-time information
GET THIS EVERY WEEK
This ranking, in your inbox — plus where the AIs disagree.
One email a week: fresh picks and the buys the five models can't agree on. First issue coming soon.
One email a week · unsubscribe any time.