Perplexity AI Review (2026): Comet, Pricing, Worth It?

5.8 percent. That's Perplexity's slice of the AI search market in early 2026, against ChatGPT's roughly 60 percent, by tracked usage. It's the underdog with the loudest reputation: the tool that made "an AI that answers with citations" famous. The catch is that in 2026, citations stopped being a product and became a checkbox. ChatGPT searches the web, Gemini has AI Mode, Claude fetches and cites pages, so the honest question this review answers isn't "does Perplexity cite its sources" but "is the citation specialist still worth paying for now that everyone does it."

A note on how we review. This is based on Perplexity's published features and pricing, its own changelog, and independent reviews and reporting as of September 2026, not on our own hands-on testing, and we earn no commission whichever way you go. AI products change fast, so check the official page before you subscribe.

What Perplexity AI is

Perplexity is an AI answer engine: a search tool that reads the live web and replies in a written answer with a citation on almost every claim. Instead of a list of blue links, you get a synthesised paragraph or two with numbered footnotes you can click to check the source. On the paid plan it routes your question through your choice of top models, and it runs a deeper "Deep Research" mode that compiles longer, sourced reports. That focus, search and synthesis with sources attached, is the whole identity. It isn't trying to be your writing or coding assistant, and that matters for whether it's worth it.

The quick verdict

• Worth it if research and fact-finding are a real part of your day, you want sources attached by default, and you'll actually open them to verify.
• Skip it if you mainly write, code, or want one tool for everything; a general chatbot that now also cites is a better fit.
• For most people, the free tier is plenty, and the honest move is to use it until you hit its limits before paying.

Free, Pro, and Max: what you get

Perplexity has a usable free tier and three paid steps above it, and they compare like this as of September 2026.

The line that matters for most readers is between free and Pro. Free covers casual questions well. Pro is about volume and depth: unlimited deeper searches, more Deep Research, and the ability to pick which model answers. If you research occasionally, the free tier rarely runs dry; if you research for a living, Pro pays for itself in time.

What Perplexity does well

The citation habit is the real draw. Because every answer footnotes its sources, checking a claim is one click away, which is exactly what you want when the facts matter and you don't trust a confident paragraph on its own. For current questions it's quick and genuinely useful, pulling together several pages into one answer faster than opening ten tabs yourself.

The Pro model picker is a quiet strength: one subscription lets you route a question through the latest models from OpenAI, Anthropic, and Google, plus Perplexity's own, so you're not locked into a single engine. Deep Research is the headline Pro feature, compiling a longer sourced report in a few minutes, and since a March 2026 update it can output that research as a presentation, spreadsheet, or dashboard. There's also a "Buy with Pro" shopping helper in the US that hunts for prices, and a free Comet browser that builds the assistant into your web browsing, which has since grown into something bigger and gets its own section below.

One more point in its favour, and a timely one: in early 2026 Perplexity dropped advertising entirely, saying ads would make people second-guess its answers, at the same moment OpenAI began adding ads to ChatGPT's free tier. For a tool whose whole pitch is trust, choosing subscriptions over ads is the right call, and worth noting when so many free tools lean the other way.

Perplexity as a computer assistant

Perplexity is not only a website any more. It now ships a browser that acts on your behalf and, on Mac, an agent that reaches into your own files. Most reviews of Perplexity still do not mention either, which is odd, because this is the biggest change to the product since Deep Research.

Comet, the browser that acts

Comet is Perplexity's AI browser, with the assistant built into the page rather than parked in a separate tab. It launched in July 2025 as a perk of the $200-a-month Max plan, went free worldwide in October 2025, reached Android that November and iOS in March 2026. The point of it is action rather than answers: it reads across your open tabs, summarises them, fills forms, and lifts data off a page into a spreadsheet. Comet Plus, which adds premium publisher content, is $5 a month and comes bundled with Pro.

Personal Computer, the agent on your Mac

Personal Computer is the bigger step, and the one worth understanding before you switch it on. Perplexity introduced it in April 2026 for Max subscribers on a waitlist, then opened it to every Mac user on 7 May 2026. It needs a Pro or Max subscription and installs as a direct download rather than through the Mac App Store. Once running it has standing access to your local files and native apps, orchestrates more than 400 connectors, and can be triggered remotely from your iPhone. It chains multi-step jobs across applications: compare two files, pull a note out of one, draft the follow-up.

The marketing does not lead with the next part. Your files stay local, but the thinking does not. The agent reads your machine and sends the work to Perplexity's servers to process. It can also change things without asking every time, renaming or resizing a file because the task implied it. Perplexity's answer to that is procedural rather than technical: CEO Aravind Srinivas says sensitive actions still require approval, every action is logged, and there is a kill switch.

Whether the trade is acceptable depends less on the feature list than on what is actually on your disk:

• If your Mac holds client work, contracts, or anything under an NDA, standing agent access plus cloud processing is a real decision rather than a settings toggle
• If it is a personal machine and the jobs are file admin, the convenience is genuine and the risk is small
• If you are on Windows or Linux, none of this applies yet, because Personal Computer is Mac only

This does not change the verdict below, because Perplexity is still bought for research. It does change who the $20 is for. Pro used to be about search volume. It is now also the entry price for an agent that can touch your desktop, and that is worth knowing before you renew.

Where it falls short

Perplexity is narrow by design, and you feel it the moment you ask for more than research. Its writing is functional at best, it has no real coding ability, and it won't generate images or hold a creative back-and-forth the way a full chatbot will. It's a research tool, so for drafting, coding, or anything creative you'll still want ChatGPT or Claude alongside it, which is why a lot of people pair the two rather than replace one with the other. Some users also report the mobile app being less stable than the web version for long documents, and the occasional rejected refund or subscription snag, which is reason enough to test the monthly plan before committing to a year.

Can you trust it?

This is the part the glowing reviews skip, and for a tool built on trust it's the most important section. Perplexity is excellent at finding sources and weaker at being right about them. It can misread a source, present a contested claim as if it were settled, and in Deep Research it has been reported to cite links that don't actually exist. The footnotes make this easy to catch if you look, and easy to miss if you don't, so the rule with Perplexity is the same as with any AI: open the sources and confirm anything that matters before you rely on it.

There's a bigger question hanging over the company, too. Through 2025 and 2026 Perplexity has been sued by a string of publishers, including the New York Times, Dow Jones, Encyclopedia Britannica, Merriam-Webster, and, in May 2026, CNN, which alleges Perplexity scraped tens of thousands of its stories. Amazon and Reddit have also taken legal action over its Comet browser and content scraping. Perplexity argues its use is fair and that it drives traffic back to publishers, and the cases are unresolved as this is written. None of that stops the product from working, but for a tool that sells itself on sourcing, how it gathers those sources is a fair thing to weigh, and worth knowing before you build a workflow around it.

Perplexity versus the chatbots that now cite too

Two years ago, citing sources set Perplexity apart. Now ChatGPT has Search, Gemini has AI Mode grounded in Google, and Claude fetches and cites web pages, so the headline feature is no longer unique. Perplexity is still the most citation-native of the bunch, sources are the default rather than an option, and the multi-model picker is genuinely handy. But the gap has narrowed enough that the choice is real.

If you want one tool that researches and also writes, codes, and chats, a general assistant is the more flexible buy. Our Perplexity vs ChatGPT comparison weighs that exact choice head to head, and the ChatGPT vs Gemini and ChatGPT vs Claude comparisons cover the other big assistants. If you want a pure research engine, Perplexity is still the cleanest, though Gemini's AI Mode runs it close on freshness, and for academic work a specialist like Consensus, which answers from peer-reviewed papers, is worth a look. For the wider toolkit, our best AI tools guide places Perplexity alongside the rest.

So, is it worth it?

For most people, the honest answer is to start free and not pay until you hit a wall, because the free tier handles casual research well. If research is a daily part of your work, Perplexity Pro at $20 a month earns its place, as long as you treat its answers as a fast, sourced first draft to verify rather than a final one. Students who research for coursework get the best deal at $10, and if you read and check more than you write, it's an easy recommendation at that price. The people who should hesitate are writers, coders, and anyone wanting a single do-everything tool, because the chatbots they'd reach for already cite their sources now. Perplexity remains the sharpest specialist in a field that has caught up with it, and whether that specialism is worth a subscription comes down to how much of your day is spent turning sources into answers. If you want to get more out of whichever tool you choose, our Gemini prompts for students and the wider library help.

What this review does not cover

This is a research-based review of features, pricing, and value, not a lab benchmark or a verdict on the lawsuits, and the picks reflect documented capabilities and reporting as of September 2026 rather than hands-on testing by us. We have not run Personal Computer against real client files, so the privacy judgement above is a reading of how it works, not a security audit. Model names, limits, prices, and legal cases all change, so confirm the current details on Perplexity's own pages, and nothing here is legal or financial advice. For how Perplexity sits against the full field, start with our best AI tools guide.

Sources

1. Perplexity: pricing and plans
2. Perplexity: changelog
3. CNN Business: CNN sues Perplexity over alleged AI copyright theft
4. Zapier: Perplexity vs. ChatGPT
5. TechCrunch: Perplexity's Personal Computer is now available to everyone on Mac (7 May 2026 general availability, Pro or Max requirement)
6. AppleInsider: Personal Computer lets AI agents access your Mac's files (local file access, approval, logging and kill switch)
7. TechCrunch: Perplexity brings Comet to Android