Perplexity Review: Answers With Sources, and One Habit to Keep
Structurally better than a chatbot for questions about the current world. The citations are an invitation to verify, not a guarantee that verification happened.
Search engines answer a question by handing you ten places where the answer might be. That was a reasonable design in 1998 and it has aged badly, because the modern search result is a wall of advertisements, a recipe blogger’s childhood memoir, and eight pages competing to rank rather than to inform.
Perplexity’s proposition is that you asked a question, so it should give you an answer — with the sources listed underneath so you can check. That is a small idea with large consequences, and it works well enough that a lot of people have quietly stopped opening Google for certain kinds of question. This review is about which kinds, and about the one habit you must keep.
Why this is different from a chatbot
Worth being precise, because “AI search” gets used for two different things.
A general assistant answers from what it learned during training, and browses when prompted. Its instinct is to answer; searching is something it does on request. That makes it excellent at reasoning and unreliable about anything current.
Perplexity inverts the default. It searches first, then writes an answer grounded in what it found, with numbered citations against the claims. The model is doing synthesis rather than recall, which is a meaningfully different job and a meaningfully different failure mode.
The consequence: for “what happened”, “what is the current price”, “what do people say about”, Perplexity is structurally better positioned. For “help me think through”, “write this”, “explain the reasoning”, a general assistant is better. Most people who find Perplexity disappointing were asking it the second kind of question.
The habit to keep: citations are provided, which is excellent, and they do not always fully support the sentence they are attached to. The synthesis occasionally drifts from the source. For anything consequential, click through. The citation is an invitation to verify, not a guarantee that verification happened.
What the tiers give you
Pricing as of August 2026, and this company ships changes quickly.
| Tier | Cost | What it means |
|---|---|---|
| Free | $0 | The most capable free tier of any comparable tool. Enough for genuine daily use. |
| Pro | $20/month, ~$200/year | Unlimited Pro Search plus a daily allowance of Deep Research runs — around twenty a day. |
| Max | $200/month | Heavy use of the more compute-intensive features. A small minority need this. |
| Enterprise | Per seat | Team controls and data handling. Priced well above the consumer tiers. |
The free tier deserves emphasis because it is unusually generous, and a lot of people would be well served never leaving it. Pro is worth it when you hit the search limits regularly or use Deep Research more than occasionally — those long-running, multi-source reports are the feature that most justifies the subscription.
Comet, and the privacy question
Perplexity also makes a browser, Comet, which has been free since late 2025 across desktop and mobile. It builds the assistant into browsing rather than making you switch to a separate tab, with shortcuts for summarising a page, generating a citation in a standard academic format, and fact-checking what is on screen. A separate small subscription unlocks premium publisher content.
The citation shortcut alone is worth knowing about if you write anything referenced. The page summariser is convenient and, like all summarisers, occasionally flattens a nuance that mattered.
The honest caveat is about privacy rather than function. An assistant integrated into your browser sees what you browse, and that is a different proposition from a search box you visit deliberately. Reviewers have consistently raised this, and it is worth reading the data handling terms before making it your default browser rather than after.
Where it falls short
Citation drift. Covered above and worth repeating, because it is the failure mode that matters. A confident answer with sources attached feels more trustworthy than a confident answer without them, and it is only more trustworthy if you actually check.
It inherits the web’s biases. If the top results on a topic are marketing pages, the synthesis will be built from marketing pages. It is better at aggregating than at judging which sources deserve weight.
Weaker as a general assistant. Writing, reasoning at length, working through your own documents — a dedicated assistant does these better. Perplexity has broadened, and search remains what it is for.
Rapid product churn. Tiers, limits and features have changed repeatedly. Good in that it improves quickly; awkward in that any specific figure ages fast.
What it does well
- Answers current questions accurately and with sources
- The most capable free tier in its category
- Deep Research produces genuinely useful cited reports
- Comet’s citation and fact-check shortcuts are practical
- Far less friction than wading through search results
What to think about first
- Citations do not always fully support the claim
- Quality depends on what ranks well on the open web
- Weaker than a dedicated assistant for writing and reasoning
- Browser integration raises real privacy questions
- Plans and limits change frequently
How it compares
| Tool | Strongest at | Weakest at | Choose it if |
|---|---|---|---|
| Perplexity Pro | Sourced answers, current information, research | Writing, working with your files | Most of your questions are about the world |
| ChatGPT Plus | Breadth, tools, code execution, files | Occasional overconfidence | You want one tool for everything |
| Navigation, local, shopping, maps | Answering questions directly | You know which site you want | |
| A library database | Source quality and depth | Speed and convenience | The work will be scrutinised |
The last row is not a joke. For academic or professional research that someone will check, a curated database beats web synthesis on source quality, and no amount of convenience compensates for that.
Who should pay
Anyone whose work involves finding things out — journalists, analysts, consultants, researchers, buyers. Deep Research alone can replace an afternoon of tab-opening.
People comparing options constantly, whether products, suppliers or approaches, where the sourced-answer format is exactly the right shape.
Students and writers who cite, where the citation shortcuts and source lists save genuine time — provided the sources get checked.
You should not pay if you mostly want help writing or thinking, if you rarely hit the free limits, or if you were hoping to stop verifying things. The tool makes verification easier; it does not remove the need for it.
Doing research?
Research usually ends in documents that need converting, merging or compressing. Our free browser-based tools handle that without uploading anything to a server.
Frequently asked questions
Is the free tier enough?
Can I trust the citations?
Is it better than ChatGPT?
Is the Comet browser safe to use?
What is Deep Research for?
Will it replace Google?
The verdict
The best tool available for questions about the current world, and the free tier is good enough that many people never need to pay. Pro earns its $20 if you use Deep Research weekly or hit the limits. Keep clicking through to sources on anything that matters — the citations are an invitation to verify rather than a substitute for it — and read the privacy terms before making Comet your default browser.
Sources and method
Pricing, tiers and Comet availability reflect Perplexity’s published information as of August 2026; this company changes its packaging frequently, so verify current figures before subscribing. The observation that citations do not always fully support the associated claim reflects consistent reporting across independent reviews rather than our own systematic testing. We have not run controlled comparisons ourselves and do not claim to have. Primary source: Perplexity’s official site.