What Perplexity is and how it differs from ChatGPT

Perplexity is an AI-powered answer engine that combines real-time web search with LLMs and shows source citations. Type a question, Perplexity sends it to a search index, fetches the most relevant pages, feeds them to a frontier model such as GPT-5, Claude Sonnet 4.5 or its in-house Sonar model, and returns a synthesised answer with footnoted sources you can click. Every paragraph is grounded in something on the open web.

That is the core difference between Perplexity and ChatGPT. ChatGPT generates text from training data and only browses the web when you explicitly enable a tool. Perplexity browses by default and treats the LLM as a writer rather than a fact source. The result is fewer hallucinations on factual queries and a built-in audit trail. The trade-off is that Perplexity is less impressive for creative writing, persistent custom personas and image generation, where ChatGPT still leads. For a head-to-head, see our Perplexity vs ChatGPT deep dive and the inverse ChatGPT vs Perplexity piece.

If you only remember one thing from this perplexity ai how to use guide, make it this: Perplexity is for questions where you would otherwise open ten Google tabs. ChatGPT is for tasks where you want one model to think out loud.

How to use Perplexity AI for the first time

Getting started takes about a minute. Open perplexity.ai in any browser, or download the iOS or Android app, and sign in with Google, Apple or email. No credit card is required. The free plan immediately gives you the search box, focus modes, Comet browser access and a small allowance of Pro Searches per day.

Type a question into the box exactly the way you would Google. The trick is to ask a real sentence rather than three keywords. Compare:

• → Google-style: "best mirrorless camera under 1500"
• → Perplexity-style: "What is the best mirrorless camera under $1,500 in 2026 for travel photography, prioritising in-body stabilisation and lightweight lenses?"

The longer prompt looks slower, but it gives Perplexity enough context to filter sources, weight criteria and structure the answer. You will see citations as small numbered chips after each sentence — hover or tap them to jump to the source. Use the follow-up box to refine, the same way you would press a journalist for clarification.

To get the most out of the free tier, turn on the Pro Search toggle for the question that actually matters and stay on Quick Search for everything else. Pro Search runs longer, fetches more sources, asks clarifying questions and uses a higher-tier model. The free plan typically allows around three Pro Searches per day, which is enough to bookmark the feature for the questions where you would normally start a research session.

That is the short version of perplexity ai how to use. The rest of this article is the long version — focus modes, Spaces, Labs, the WhatsApp bot, Comet, the Assistant, the API and the daily routine that makes Perplexity actually replace Google.

Focus modes — Web, Academic, Social and more

Focus modes tell Perplexity which corner of the web to search in. They sit just under the search box on web and inside the prompt sheet on mobile. Choosing the right focus mode changes the source mix and, in many cases, the model behaviour.

• → Web: the default. Searches the open index. Use it when you are not sure where the answer lives.
• → Academic: restricts to peer-reviewed papers, preprints and reputable journals. Use it for literature reviews, scientific claims, citations for a thesis.
• → Social: pulls from Reddit, X and other social discussions. Useful for sentiment, real-world experiences, troubleshooting threads and travel tips.
• → Video: indexes YouTube and other video sources. Pair it with a question like "show me the best 10-minute walkthroughs of using Cursor for the first time".
• → Writing: turns off web search and uses the LLM directly. Best for drafting emails, outlining, brainstorming, copy editing.
• → Math: activates a code-interpreter style mode that runs calculations and step-by-step proofs. Use it for unit conversions, finance math, statistics.
• → Finance: on Pro, pulls live prices, filings and analyst commentary. Use it for ticker-level questions and market briefings.

The most underused focus mode is Writing. Many users assume Perplexity always searches the web and avoid it for drafting. In Writing mode it behaves like a clean ChatGPT, so you can compose long emails or meeting summaries without citations cluttering the output.

How to use Perplexity AI effectively — prompts that work

Most people who say Perplexity "did not work" sent a one-line query and accepted the first answer. Here is how to use Perplexity AI effectively, with patterns that consistently produce better answers.

Constrain the answer format

Tell Perplexity exactly what shape the answer should take: "Give me a 5-row markdown table with columns: tool, free plan, starting price, best for, source". Format constraints lower hallucinations because the model has fewer ways to drift.

Constrain the time window

Add a date filter to the prompt itself: "Answer using only sources published after January 2026." Perplexity passes the constraint to the search layer, so you stop getting answers that quote 2023 blog posts as if they were current.

Ask for sources by tier

For research-grade answers, write something like: "Prefer primary sources — official documentation, SEC filings, peer-reviewed papers — and clearly mark which sources are blog posts or vendor pages." This is the closest thing Perplexity has to a quality knob.

Use multi-turn refinement

Treat each search as a starting point. Ask one broad question, scan the citations, then drill in: "Now expand step 3 — what does the regulator's draft guidance from March 2026 actually require, and where does it diverge from the EU AI Act?" Each follow-up keeps the source set from the original thread, which is why the chain matters.

Pin a model when it matters

On Pro, click the model selector and choose deliberately. GPT-5 and GPT-5.2 are strong all-rounders, Claude Sonnet 4.5 is the best for nuanced reading and long-form writing, and Sonar Reasoning is the cheapest fast option. For coding questions, Claude tends to win. For finance and math, GPT-5 with Math focus is hard to beat.

For more prompt patterns see our best AI search engines 2026 roundup, which compares Perplexity to Google's AI Overviews, You.com and ChatGPT search side by side.

How to use Perplexity AI in your daily routine

The compound benefit of Perplexity comes from making it your default for small information tasks. Here is a realistic week of how to use Perplexity AI in daily routine work, drawn from how the ToolChase team actually uses it.

• → Morning briefing: ask "Brief me on what happened in AI between 06:00 yesterday and 06:00 today, grouped by company, with a one-sentence so-what for each." Save the prompt as a Space (see below).
• → Inbox triage: paste the body of a long email and ask "Summarise in three bullets, flag any deadlines and draft a polite two-sentence reply."
• → Quick fact-checks: drop a claim into Perplexity instead of opening a Google tab. The citations are the point.
• → Mid-day research: any time you would have opened ten tabs, run a Pro Search instead. Then keep the thread open and ask follow-ups as new questions appear.
• → Code-adjacent tasks: for "what is the API limit on plan X" or "what is the latest version of library Y", Perplexity is faster than reading docs.
• → Travel and shopping: Comet's agentic Assistant compares hotel rates, flights, return policies and warranty terms in tabs you already have open.
• → Evening capture: at the end of the day, ask Perplexity to write a 5-bullet summary of the threads you opened. It will pull the source set forward.

The pattern is "ask, do not open". If you train yourself to ask Perplexity before opening a search engine, the productivity gain in week two is meaningful. Pair it with a paid plan if you do this professionally — see our Is Perplexity Pro worth it analysis.

How to use Perplexity Spaces — collaborative project context

Spaces are Perplexity's answer to ChatGPT Projects and Claude Projects. A Space is a persistent workspace that holds three things: standing instructions for the AI, uploaded files, and thread history. Every search inside a Space inherits all three. This is how to use Perplexity Spaces well in practice.

Open the left navigation, click Create a Space, give it a name like "Q2 competitor research" and write standing instructions in plain English. Useful patterns:

• → "Always cite peer-reviewed sources before blogs, and label vendor pages as such."
• → "I am a B2B SaaS founder researching the AI sales-engagement category. Default to founder-relevant framing — pricing, GTM, churn, ICP."
• → "Respond in UK English. Use metric units. Format dates as DD MMM YYYY."
• → "Treat anything I upload as authoritative. If web sources contradict an uploaded file, flag the contradiction."

Then upload the documents that anchor the Space. PDFs, Word documents, Excel sheets, Markdown notes, even images all work. Pro users get larger upload allowances and more files per Space than the free tier. Once instructions and files are in place, every thread you start in that Space behaves as if you had pasted everything into the prompt — but the context never gets pruned and you can come back weeks later.

Spaces also support collaboration. Invite teammates as viewers or contributors and they can run their own threads against the same context. This is genuinely better than Slack threads for any topic that runs longer than a week. Use a Space for each client, each long-running research project, and each recurring report — for example, the morning briefing prompt above lives in a "Daily AI" Space so the model knows to skip topics already covered yesterday.

How to use Perplexity Labs — long-form output mode

Labs is Perplexity's "go beyond a chat answer" mode. In the product it is now surfaced under Create files and apps in the + menu, but most people still call it Labs. A Labs run can take five to fifteen minutes and produces something more like a deliverable than a reply: a multi-page PDF report, a CSV with filled-in data, a slide deck, a chart, or a small interactive web app.

Here is how to use Perplexity Labs without wasting runs:

1. Pick a deliverable, not a question. "Build a market map of the AI customer-support category with 15 vendors, columns for pricing, target ICP, and key integrations" works. "Tell me about AI customer support" does not.
2. Specify the output format: PDF, slides, dashboard, table, web app.
3. Tell Labs the audience: "Internal exec audience, no marketing fluff, cite every claim."
4. Provide constraints: word count, slide count, style.
5. Iterate. Labs is best used as a first-draft generator, not a one-shot oracle.

Useful Labs prompts to keep in your back pocket: "Generate a competitive teardown of [product] including pricing, positioning, weaknesses and 3 angle-of-attack ideas, exported as a PDF report." "Build an interactive dashboard that compares EV charging speed by manufacturer using public data, deployed as a web app." "Produce a 12-slide pitch deck on [topic] with a clean title slide, problem, solution, market, competitive landscape, traction and ask."

Labs is included with Pro and Max. Pro users have monthly Labs query limits, while Max users get a far larger allowance. Treat each run like a paid research request and you will get more out of it.

How to use the Perplexity AI Assistant — voice, Android and iOS

The Perplexity Assistant turns the search engine into a phone-grade voice assistant. On Android you can set Perplexity as the default assistant in system settings and trigger it the way you used to trigger Google Assistant. On iOS, the Voice Assistant lives inside the Perplexity app and on the lock screen via the Action Button or Control Center shortcut. On Samsung Galaxy S26, Perplexity is wired in at the OS level and answers to "Hey Plex" or a long-press of the side key.

What makes the Assistant different from Siri or Google Assistant is that it can complete multi-step tasks rather than just answer trivia. Sample voice prompts that work today:

• → "Find me three highly rated Italian restaurants near me that are open after 9pm and book the closest one for two at 9:30."
• → "Send a text to Maya saying I am running 15 minutes late."
• → "Draft a summary of my last email from the founder of [company] and tell me the three things I should reply to."
• → "Compare the two flights I screenshotted and pick the cheaper one with under one hour layover."

The Assistant runs in the foreground, narrates what it is doing and asks for confirmation before transactional steps like sending or paying. Voice mode currently uses an OpenAI realtime model under the hood, which is why interruptions feel natural and the assistant can be cut off mid-sentence. This is the cleanest answer to how to use perplexity ai assistant on a phone in 2026.

Comet browser — Perplexity's agentic browser

Comet is Perplexity's full Chromium-based browser. It looks like Chrome and imports your bookmarks, but the address bar is a Perplexity prompt and there is a permanent Assistant sidebar on the right of every tab. Comet originally cost $200 per month and was Max-only. In March 2026 Perplexity dropped the paywall and shipped Comet free on Mac, Windows, iOS and Android.

Comet does three things no traditional browser does:

• → Reads the page you are on. Open any article, ask "summarise this in 5 bullets and tell me what is missing" and the sidebar answers using the page text plus a fresh web search for what is missing.
• → Acts across tabs. Have ten travel tabs open and ask Comet "rank these flights cheapest to most expensive and tell me which ones include a checked bag". The Assistant reads them all.
• → Completes tasks agentically. Tell Comet "fill out the visa form in tab 4 with my passport details from tab 2, then upload the photo from my downloads folder". It clicks, types and submits with confirmation gates.

Comet is free for everyone. Comet Plus, a $5 per month add-on, unlocks premium publisher access from media partners and is bundled with Pro and Max. If you only ever try one Perplexity feature beyond the search box, make it Comet — it is the closest thing to a useful agentic browser shipped by anyone in 2026.

How to use Perplexity on WhatsApp — the WhatsApp bot setup

Perplexity ships a WhatsApp bot that turns the world's most-used messenger into an AI search box. There is no app to install and no account to create. This is how to use Perplexity AI on WhatsApp in three steps.

1. Save +1 (833) 436-3285 in your phone contacts as "Perplexity". Or skip the save and tap wa.me/18334363285, which opens a chat directly.
2. Open WhatsApp, find the contact, send "hi" and the bot replies with a short welcome.
3. Ask anything. The bot returns a Perplexity-style answer with source links inline. You can also send images for visual question answering and forward links for instant summaries.

This is the cleanest path to how to use perplexity ai on whatsapp if you live in WhatsApp anyway. It is also the answer to how to use perplexity ai assistant on whatsapp, because the WhatsApp bot is effectively the Assistant in chat form: it can answer, summarise, generate images and stream voice notes back. The bot is global, free and account-less, but availability can change because Meta restricts third-party AI bots through its WhatsApp Business API rules. If the bot stops responding in your region, the desktop and mobile apps still work normally. For a deeper walkthrough including the how to use perplexity ai whatsapp bot tips for groups and broadcasts, the Perplexity Help Center has the canonical reference.

How to use the Perplexity API — quick start

Developers can call Perplexity's models directly via the API at docs.perplexity.ai. The headline reason to pick it over a generic LLM API is that Sonar models return citations in the response, which removes most of the work of building retrieval-augmented question answering. Here is the short version of how to use the Perplexity API.

1. Sign in at perplexity.ai, open Settings → API, add a payment method and generate a key.
2. Choose an endpoint. The chat completions endpoint at api.perplexity.ai/chat/completions is the workhorse. The search endpoint returns ranked URLs without LLM synthesis. The v1/agent endpoint exposes the agentic research pipeline used by Labs.
3. Pick a Sonar model. sonar is the cheap fast option, sonar-pro trades latency for citation quality, and sonar-reasoning is the long-context reasoning variant. You can also route to GPT-5 family models via the same API.
4. Send a request with a Bearer token in the Authorization header.

A minimal cURL call looks like this:

curl https://api.perplexity.ai/chat/completions \
  -H "Authorization: Bearer $PERPLEXITY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "sonar-pro",
    "messages": [
      {"role": "user", "content": "What were the most-cited AI papers published in May 2026?"}
    ]
  }'

The response includes the assistant message and a citations array of URLs the model used. From there it is a normal API: stream responses with server-sent events, set search_recency_filter to bound the time window, pass a search_domain_filter to restrict to allow-listed sites, or set return_images for multimedia answers. Pricing is per million tokens with separate rates for input, output and search calls — check the docs page for the current numbers because they change. Pro accounts come with monthly API credits, which is the cheapest way to prototype.

Free vs Pro — what you unlock with $20/mo

Perplexity has five tiers in 2026: Free, Pro at $20 per month or $200 per year, Max at $200 per month, Education at $10 per month for verified students and educators, and Enterprise from $40 per seat per month. For most readers the choice is Free vs Pro. Here is the difference in plain English.

The fastest way to know if you want Pro: open the model selector menu while logged in. If you find yourself wishing you could pin Claude Sonnet 4.5 for a writing question and GPT-5 for a coding one, you want Pro. If you only ever ask quick factual questions, the free tier is fine. We do a deeper cost-benefit analysis in Is Perplexity Pro worth it and a full editorial review in our Perplexity AI tool review.

Two more details that matter. First, annual billing at $200 saves about 17 percent. Second, several carriers and partners run promos that bundle a free year of Pro — check before you pay. Verified pricing is updated on perplexity.ai/pro.

Putting it all together

The fastest mental model for Perplexity in 2026 is: search box for facts, Spaces for projects, Labs for deliverables, Assistant for hands-free, Comet for agentic browsing, WhatsApp for couch-mode questions, API for product builders. Free covers most casual users; Pro pays for itself the first time you replace a research session with a Pro Search.

If you came here looking for how to use perplexity end to end, that is the playbook. Set Perplexity as your default search engine, save the WhatsApp number, install Comet, set up two or three Spaces and run one Labs project on a topic you actually care about. After two weeks the Google tab habit breaks on its own.