★★★★★Trustpilot · October 2026
A user says an Enterprise Max subscription was terminated four days into a paid period with no notice or refund and the ticket went unanswered.
We tested Perplexity for 7 days, from October 16 to October 22, 2026, giving it the same 30 everyday tasks we use for every AI chatbot we review. It got all 30 right, made no factual errors we could find, and three judges who didn't know which tool wrote what gave its answers 5 out of 5. That puts it second of the 12 assistants we tested, with an overall score of 5 out of 5.
What sets it apart is that every answer comes with links to the web pages behind it. We checked those links by hand, and 29 of the 30 held up. Not one task needed correcting.
Perplexity suits researchers, analysts, students and anyone who needs to see where an answer came from. The free plan covers light use, and Pro costs $20 a month. Just know that many users complain about billing and cancelling, so read the pricing section before you subscribe.
How we scored it
Perplexity scored 5 out of 5 on all five parts, so the parts were even. Nothing in our test held one part back more than another.
Output quality5.0
Accuracy5.0
Ease of use5.0
Features and limits5.0
Value for money5.0
Overall score: 5 / 5, the average of the five parts, rounded to the nearest half point.
- Output quality (5/5): Three blind judges gave its answers 5 out of 5, without knowing which tool wrote them.
- Accuracy (5/5): It got all 30 tasks right and we found 0.0 factual errors per 1,000 words.
- Ease of use (5/5): You type or speak a question and get a sourced answer, and 0 of our 30 tasks needed correcting.
- Features and limits (5/5): Cited answers, Deep Research, model choice and the free Comet browser cover research needs well, though paid plans still have limits.
- Value for money (5/5): The free plan covers cited search, and Pro at $20 a month adds Deep Research and model choice for results that matched the best in our test.
Best of 2026
See the ranking →
The best AI chatbots and assistants
Before you pick Perplexity, see how all 12 tools ranked in the same test.
Perplexity in plain terms
Perplexity is an AI answer engine. You ask a question, it searches the web, and it writes a short answer with links to the sources it used. Think of it as a search engine that reads the pages for you, but still shows you where the facts came from.
It runs on the web and in apps for iPhone and Android. Pro adds Deep Research for longer reports, a choice between several leading AI models and higher limits. Max adds the most use of Perplexity Computer, an agent that carries out tasks for you, and the highest limits. The company also makes Comet, an AI web browser that has been free since October 2025.
That makes it different from general chatbots built mainly for writing and coding. In our broader look at AI chat tools, Perplexity is the one that leans most toward research and fact-finding. Perplexity AI, Inc. was founded in 2022 by Aravind Srinivas, Denis Yarats, Johnny Ho and Andy Konwinski.
What happens after you sign up
Using Perplexity feels like using a search engine, but you get an answer instead of a list of links.
1. Ask a question. Type or speak your question on the website or in the app.
2. Perplexity searches the web. It looks up sources and writes an answer with numbered links to them, so you can check each point.
3. Follow up or go deeper. Ask follow-up questions in the same thread, or run Deep Research on a paid plan for a longer report.
4. Pick a model if you want. On Pro you can choose which of several top AI models writes the answer.
What you get
Perplexity's features centre on finding and checking information. The free plan covers the basics, and paid plans add depth and higher limits.
Cited answers
Every answer links to the sources behind it, on every plan. In our test, 29 of 30 citations held up when we checked them.
Deep Research
On Pro and Max, it builds longer research reports from many sources, which saves hours of reading.
Model selection
Pro and Max let you choose which of several leading AI models answers, so you can compare them without separate subscriptions.
Perplexity Computer
An agent that carries out tasks for you. Pro gets expanded access and Max gets 10,000 Computer credits a month.
Finance data
Pro lists data from PitchBook, Statista and S&P Capital IQ, which helps if you research companies or markets.
Who gets the most out of Perplexity
Perplexity is at its best when you need facts you can check. It's less suited to people who mainly want long creative writing or heavy coding help, or who expect unlimited use.
A good fit for
- Researchers, analysts and students who need cited answers
- People who want to compare several top AI models on one plan
- Anyone who researches companies or markets
- People curious to try a free AI browser
Not a good fit for
- Writers or developers who mainly want creative writing or coding help
- Heavy users who expect unlimited use on Pro
- Anyone uneasy about publisher disputes over how it gathers content
- Buyers who want a clear, easy refund policy
Perplexity vs ChatGPT, Claude and Google Search
Perplexity sits near the very top of our lineup. ChatGPT came first, also with 5 out of 5, and is the better pick if you want one assistant for everything, with paid plans from $8 a month. Claude scored 4.5 and is stronger for long documents and coding. Our best AI chatbots list explains the full ranking.
Google Search gives you links to read yourself, while Perplexity writes the answer and shows its sources. It saves a lot of time over research by hand, but open the sources for anything important.
| Option | Price | How it compares |
|---|---|---|
| Perplexity | Free plan; Pro $20 a month | Searches the web and writes an answer with links to its sources. |
| ChatGPT | Free plan; Go $8 a month | General AI chatbot with the broadest feature set and app ecosystem. |
| Claude | Free plan; Pro $20 a month | AI assistant that is strong at writing, long documents and coding. |
| Google Search | Free | Lists links to pages, which you read and weigh yourself. |
Our 7-day Perplexity test
We tested Perplexity from October 16 to October 22, 2026, 7 days in all. Every assistant in this group gets the same 30 everyday tasks, pasted in exactly as written, on the plan most buyers would pick and with no custom instructions. The tasks include research questions with sources, emails, a summary of a 20-page PDF, a spreadsheet question and a coding fix. The full set-up is in how we test.
We marked each answer against a written answer key. Perplexity got all 30 tasks correct and complete. In plain terms, it didn't miss a single one, and none of its answers needed fixing before we could use them.
Next we counted factual errors across everything it wrote. We found 0.0 errors per 1,000 words. For a tool built around research, that's the result that matters most: you can lean on what it tells you.
Then three judges rated the answers blind, without knowing which assistant wrote what. They gave Perplexity 5 out of 5 for quality, so its answers weren't just correct but clear and useful.
Because Perplexity shows its sources, we also opened its citations by hand. 29 of the 30 checked out. That one miss is a good reminder: the links make checking easy, but they only help if you actually click them.
| What we measured | Result | What it means |
|---|---|---|
| Correct and complete answers out of 30, checked against an answer key | Tasks correct: 30/30 | Every task came back right and complete, so you can expect usable answers for everyday work. |
| Factual errors found per 1,000 words | 0.0 errors/1,000 words | We found no factual mistakes, which is what you want from a research tool. |
| Blind quality score from three judges | 5.0/5; three blind judges | Judges who didn't know the source rated its answers top marks for clarity and usefulness. |
Perplexity did everything we asked. It got all 30 tasks right, made no factual errors we could find, earned top marks from blind judges, and 29 of its 30 citations checked out. The weak spots weren't in the answers but around them: limits on paid plans and many user complaints about billing.
Tested October 16, 2026 to October 22, 2026.
The good and the bad from our test
Perfect score on our task set
Perplexity got all 30 tasks right and we found 0.0 factual errors per 1,000 words. Not one answer needed correcting.
Sources you can check
Every answer links to its sources, and 29 of 30 held up when we checked them by hand. That makes it easy to verify anything before you rely on it.
Strong free plan
The free plan gives you cited search with basic models for limited daily use. That's enough to see whether it suits you before paying $20 a month for Pro.
One citation in 30 didn't hold up
29 of 30 citations checked out, so one didn't. It's a small miss, but it means you should still open the source for anything important.
Many billing and cancellation complaints
On Trustpilot, checked November 2, 2026, Perplexity has 1.4 out of 5 from 851 reviews, with 83% one-star. Most complaints are about billing, credits and cancelling.
Paid plans aren't unlimited
Pro includes 4,000 bonus credits and higher limits, not unlimited use. Some users report being charged credits after hitting weekly limits mid-project.
Pros
- All 30 test tasks answered correctly
- 0.0 factual errors per 1,000 words in our test
- Answers link to their sources
- Useful free plan for light research
- Choice of several top AI models on Pro
Cons
- 1 of 30 citations didn't hold up
- Trustpilot rating of 1.4 out of 5
- Credits and limits apply on paid plans
- No refund policy we could confirm
- Max is expensive at $200 a month
How much Perplexity costs
Perplexity has a free plan and two paid plans. Free gives you cited search and basic models for limited daily use. Pro costs $20 a month billed monthly, or $17 a month billed yearly. Max costs $200 a month billed monthly, or $167 a month billed yearly. Business plans are custom, through a sales demo.
Most buyers should start on the free plan and move to Pro if they hit the daily limits or want Deep Research and model choice. Max is for heavy users who want the highest limits and the most Perplexity Computer use. Prices were checked on November 2, 2026.
Free
$0
a month
- Cited search answers
- Basic models for limited daily use
Pro
$20
a month ($17 a month billed yearly)
- Deep Research and model selection
- 4,000 bonus credits
- Expanded Perplexity Computer access
Max
$200
a month ($167 a month billed yearly)
- 35,000 bonus credits
- 10,000 monthly Computer credits
- Highest limits and priority access to new features
The free plan is a lasting plan, not a trial, and covers cited search with basic models for limited daily use. Perplexity doesn't state a free trial of its paid plans.
Paying yearly lowers Pro to $17 a month and Max to $167 a month. Enterprise pricing is custom, and the pricing page points to a sales demo with no price shown.
We couldn't confirm a refund policy on Perplexity's official site. User reports are mixed, so don't count on getting money back if you change your mind.
Prices checked November 2, 2026. Prices can change, so check before you buy.
Is Perplexity safe and reliable to use?
We found no reports of a major data breach. We couldn't confirm how Perplexity uses or keeps your data, because its privacy page wasn't available when we checked, so avoid pasting sensitive business information until you've read its current policy.
Perplexity faces several legal disputes with publishers over how it uses their content, including Dow Jones and the New York Post, Britannica and Merriam-Webster, Reddit, CNN and Japanese publishers. In August 2025 Cloudflare alleged it used undeclared crawlers that ignored website rules, which Perplexity denied. The BBC also threatened legal action in June 2025.
None of this affects your account directly, but it matters if your business cares about how AI tools treat publishers. The more practical risk is billing: Trustpilot users report subscriptions ended without notice or refund and cancellation that's hard to find.
Open the cited sources before relying on an answer, and cancel your subscription before deleting your account.
What Perplexity users say
Users are sharply split depending on where you look. On the App Store, checked November 2, 2026, the Perplexity app has 4.8 out of 5 from 521,000 ratings. On Trustpilot, checked the same day, it has 1.4 out of 5 from 851 reviews, labelled Bad, with 83% giving 1 star and 8% giving 5 stars.
The praise is about answer quality: people like the cited answers, and some call it the best AI for research. The complaints are mostly about money and access: cancellation that's hard to find, credits charged after limits, subscriptions ended without a refund and limited free use. In October 2026, one user said a paid plan ended a few days in with no notice or refund, while another praised a near-instant refund.
What users praise
- Cited answers
- Seen by some as the best AI for research
- Fast refund reported by one user
- Easy upgrade
What users complain about
- Cancellation hard to find
- Credits charged after limits
- Subscriptions terminated without refund
- Limited free use
Recent user reviews, in short
Summarised in our own words, not quoted. Reviewer names are left out.
★★★★★Trustpilot · October 2026
A reviewer says credits were charged after a weekly limit was hit mid-project and table errors went unrefunded.
★★★★★Trustpilot · October 2026
A user calls it the best AI available but notes Pro lacks unlimited credits; pausing and changing the plan was easy.
★★★★★Trustpilot · October 2026
A user says they were refunded almost immediately and praises the support.
Our test lines up with the praise. Perplexity's answers were accurate, well sourced and highly rated by blind judges. We didn't test billing or cancellation, so the Trustpilot complaints are worth taking seriously: keep track of your plan's limits and renewal date.
Ratings checked November 2, 2026.
Perplexity alternatives worth a look
We tested 12 AI assistants the same way. These five scored highest after Perplexity, or level with it.
ChatGPT
ChatGPT came first with 5 out of 5. Pick it if you want one general-purpose assistant with the broadest feature set and app ecosystem.
Microsoft Copilot
Microsoft Copilot scored 5 out of 5. It suits people already on Microsoft 365, Windows or Edge who want an assistant in their Office apps.
Grok
Grok scored 5 out of 5. It suits people who spend their time on X and want an assistant with image and video tools.
Claude
Claude scored 4.5 out of 5. Pick it if you mainly work with long documents, writing or coding rather than web research.
DeepSeek
DeepSeek scored 4.5 out of 5 and is free. It suits cost-conscious users who are comfortable with their data being hosted in China.
5/5
Should you use Perplexity?
We recommend Perplexity for anyone whose work depends on accurate, sourced answers. It got all 30 of our tasks right, we found no factual errors, blind judges gave it top marks, and 29 of its 30 citations checked out. That's why it scored 5 out of 5 and came second of the 12 assistants we tested.
The catches are around the edges. Paid plans have limits and credits rather than unlimited use, we couldn't confirm a refund policy, and many Trustpilot users report billing and cancellation trouble. Start on the free plan, move to Pro at $20 a month if you need Deep Research, and keep an eye on your renewal date.
Perplexity suits anyone who needs accurate answers with sources; people who mainly want writing or coding help may prefer ChatGPT or Claude.
Still deciding?
Compare AI chatbots and assistants →
See how Perplexity compares
Every tool in our AI chatbots and assistants test got the same jobs. Our ranking puts the results side by side.
Disagree with our review?
Make your case
If you use Perplexity, or work there, and think we got something wrong, email us. Tell us what you saw and attach screenshots if you have them. We'll take another look and update the review if it changes our findings.
review.now.team@gmail.com
or open your email app
Frequently asked questions
How much does Perplexity cost?
Perplexity has a free plan for limited daily use. Pro costs $20 a month billed monthly, or $17 a month billed yearly, and adds Deep Research and model choice. Max costs $200 a month billed monthly, or $167 a month billed yearly, with the highest limits. We checked these prices on November 2, 2026.
How did you test Perplexity?
We gave Perplexity the same 30 everyday tasks we give every AI assistant, over 7 days from October 16 to October 22, 2026. We marked answers against a written key, counted factual errors, had three blind judges rate quality, and checked its citations by hand. It got all 30 tasks right and 29 of 30 citations held up.
Is Perplexity free?
Yes, there's a free plan. It gives you cited search answers with basic models for limited daily use. For Deep Research, a choice of top models and higher limits, you need Pro at $20 a month. The Comet browser from Perplexity is also free.
Is Perplexity better than ChatGPT?
It depends on the job. Both scored 5 out of 5 in our test, with ChatGPT ranked first and Perplexity second. Perplexity centres on cited answers and research, while ChatGPT is a broader chatbot with a larger user base and more features. If checking sources matters most, Perplexity is the better fit.
Can I trust Perplexity's sources?
Mostly, yes. In our test, 29 of its 30 citations held up when we opened them, and we found 0.0 factual errors per 1,000 words. That one miss shows why you should still open the source for anything important, especially before you share or act on it.
Does Perplexity give refunds?
We couldn't confirm a standard refund policy on Perplexity's official site. User experiences vary: some Trustpilot users say they were refunded quickly, while others say their subscription ended without a refund. Try the free plan first so you know Pro is worth it before you pay.
Written by
Liam Foster
If there's a strange new AI tool making its way around the internet, I've probably already bookmarked it. My favorite part is figuring out what these platforms can actually do.