Perplexity works as a research tool by pairing an AI-written answer with the actual sources it pulled from, so you get numbered citations to check instead of a confident guess. This guide covers how to use Perplexity for research end to end — the search box, Focus modes, Deep Research, and Spaces — with the exact steps and prompts that got me useful, checkable answers fastest.
Short answer: Go to perplexity.ai or the app, type your question, and read the numbered citations under the answer. Switch Focus mode to Academic or Writing for a narrower source type, and turn on Deep Research when a question needs a multi-source report instead of one paragraph. Free covers most casual research; Pro ($20/month) removes the daily caps and adds Spaces.
In my testing, I ran the same set of research questions through Perplexity's default search, then again with Deep Research turned on, to see where the extra step actually earned its time cost. It did, but only for questions with more than one sub-part — a plain fact-check didn't need it. Here's what happens at each step, when Deep Research is worth the wait, and the mistakes that make Perplexity look less reliable than it is.
What you'll need
Nothing to install to get started. Go to perplexity.ai in any browser, or install the mobile app on iOS or Android — a free account (email or Google/Apple sign-in) gets you real search, Deep Research, and citations with no credit card required. If you want to attach files, have your PDFs, spreadsheets, or images ready; free accounts can upload files, though Pro raises the volume you can keep in a persistent Space. If your goal is an AI-native browser rather than a chat tab, Perplexity's Comet browser is free for anyone now, on Mac, Windows, iOS, and Android, after Perplexity dropped its $200/month Max-only paywall on October 2, 2025, per CNBC’s reporting at the time.
Step-by-step: How to use Perplexity for research
1. Ask a plain question first
Type your question into the search box on perplexity.ai or in the app and press enter. You don't need special syntax — a normal, specific question ("what did the FDA say about X in 2025" beats "tell me about X") gets a sourced answer in a few seconds, with the model's default set to Perplexity's own Sonar unless you change it.
2. Pick a Focus mode when the source type matters
Click the Focus selector above the search box to restrict results to a specific source type: Academic pulls from peer-reviewed papers and preprints, Writing turns off web search for pure drafting, Video indexes YouTube content down to the timestamp, and Social scans Reddit and X for real public sentiment instead of official pages. Leaving it on the default "All" is fine for general questions; switch it the moment a topic is scientific, YouTube-specific, or opinion-based.
3. Turn on Deep Research for anything with sub-questions
For a question that actually has parts — "compare these three tools," "summarize the last two years of research on X" — switch on Deep Research instead of a normal search. It runs a multi-step process: breaking your question apart, searching and reading dozens of sources, then writing a structured report with citations for each claim. It takes longer, often a couple of minutes, so save it for questions a single search would answer shallowly.
4. Save recurring research into a Space
Click Spaces in the sidebar to create a persistent workspace that bundles a system prompt, uploaded files, and prior conversations around one topic — a client, a class, an ongoing project. Every new thread inside that Space keeps that context automatically, so you stop re-explaining the same background every time you open a new chat.
5. Attach your own documents when the question is about them
Click the attachment icon to upload a PDF, spreadsheet, or image directly into your question. Perplexity reads the file as part of that turn, which is faster and more accurate than pasting text manually, and it's the right move any time you're asking about a specific document rather than the open web.
6. Switch models manually if you're on Pro or Max
Paid plans let you pick which underlying model answers a given query from a dropdown, instead of Perplexity choosing Sonar automatically. Use that when a question needs harder reasoning or long-context handling than the default model gives you; for most day-to-day lookups, the default is genuinely fine.
Example prompts you can copy
The gap between a flat answer and a genuinely useful one comes down to how much structure you put in the prompt, not the topic itself. A few I reuse:
- "Compare [option A] and [option B] on price, the single biggest limitation, and who each is actually built for. Cite your two most recent sources."
- "Run Deep Research: summarize the last two years of research on [topic], and flag anything sourced from a non-peer-reviewed site."
- "I'm deciding between [option A] and [option B] for [specific use case]. Give me a one-line recommendation first, then the supporting detail below it."
- "Summarize this PDF's methodology and findings in five bullet points, then list any numbers I should verify against the original document: [attach file]."
- "Explain [topic] the way you would to someone who already understands [related topic] but hasn't seen this specific study before."
Each one states the comparison, the format, or the constraint Perplexity should aim for — a bare topic name almost always gets a bare, generic answer back.
Common mistakes to avoid
The one I made first: running a plain search for a question that had three sub-parts, then being annoyed the answer felt shallow — that's a Deep Research job, not a single search. Second, trusting a confident-sounding number without opening the citation underneath it; Perplexity is good about attaching sources, but a source being cited isn't the same as the claim being read correctly, so I check anything I'll actually repeat. Third, skipping Academic Focus mode on a scientific claim and then being surprised the top citation is a random blog post instead of a paper. Fourth, starting a big recurring research project in a regular chat instead of a Space, then losing the thread across a dozen unrelated conversations. And fifth, uploading an entire 60-page report when only one section matters — point Perplexity at the relevant part instead, the same habit worth having with any AI tool that reads documents.
Free vs. Pro vs. Max, at a glance
Free is a real research tool, not a stripped demo — the gap to paid plans is mostly about daily caps, not core features.
| Plan | Price | Deep Research | Spaces & files | Best for |
|---|---|---|---|---|
| Free | $0 | Limited daily queries | Basic uploads, no persistent Space | Casual fact-checks, occasional research |
| Pro | $20/mo (~$200/yr annual) | Much higher daily allowance | Full Spaces, larger file uploads | Regular research, writing, recurring projects |
| Max | $200/mo (~$2,000/yr annual) | Highest tier, priority | Everything in Pro plus top-tier model access | Heavy daily use, frontier-model comparison |
Perplexity doesn't publish one fixed "X queries per day" number for the free tier the way a price page does — the exact daily cap has moved more than once this year, so treat any specific figure you see elsewhere as a snapshot, not a guarantee. What's stable: Pro and Max both remove the tightest caps and add Spaces, and Comet, Perplexity's AI browser, is free for everyone on any plan after the company dropped its $200/month Max-only price in October 2025, according to eesel’s pricing timeline.
Tools that make this easier
Perplexity isn't the only place worth doing research, and it isn't always the fastest for every kind of task. If you mostly want a chat assistant with web search baked into a familiar interface, how to use ChatGPT’s search engine covers the closest direct comparison. Google's own Deep Research equivalent lives inside Gemini, and I put real timing numbers on it in how to use Gemini AI — worth reading if you're deciding which deep-research tool to actually pay for. If your research is really the first step before writing, not the end goal, how to use Claude AI is the strongest option I've tested for turning a pile of sourced notes into a clean draft. When the research question is specifically about ranking or visibility rather than a general topic, best AI tool for SEO optimization and what is the best AI tool for increasing visibility cover that narrower case directly. And if the output of your research is going to be published content at volume, my ranked best AI writing tools guide covers what to hand it off to next.
My take
Perplexity earns its keep on questions that have more than one part — comparisons, literature summaries, anything where you'd otherwise open eight tabs yourself. For a single quick fact, it's not meaningfully better than any other AI search box, and Deep Research is overkill for those. The free plan is genuinely usable for casual lookups; Pro's $20/month mostly buys back the daily caps and adds Spaces, which is worth it the moment you're researching the same subject across more than one sitting. The habit that matters most isn't which plan you're on — it's opening the citation before you repeat the claim.
Frequently Asked Questions
Is Perplexity free to use for research?
Yes. The free plan includes real search, citations, and a limited daily allowance of Deep Research queries with no credit card required. Pro ($20/month) and Max ($200/month) raise those daily limits and add Spaces and higher-tier model access.
How long does it take to learn how to use Perplexity for research?
A few minutes for a basic search — type a question, read the citations. Learning when to reach for Deep Research versus a plain search, and when to switch Focus mode, usually takes a handful of real research sessions to feel automatic.
What is the easiest way to get good results from Perplexity?
Give it a specific, multi-part question instead of a bare topic, pick the right Focus mode for the source type you need, and open the citation on anything you plan to repeat as fact rather than trusting the summary alone.
Is Perplexity's Comet browser free now?
Yes. Comet launched exclusively for Max subscribers at $200/month in mid-2025, then Perplexity made it free worldwide on Mac, Windows, iOS, and Android starting October 2, 2025, per CNBC’s coverage.
Does Deep Research replace a normal search?
No — use it only for questions with real sub-parts, like a comparison or a multi-source summary. For a single fact or quick lookup, a normal search is faster and just as accurate, since Deep Research's extra minute or two of processing doesn't add value to a one-part question.