An answer engine for the curious. Ask anything and get instant, source-backed answers across every model, every device, and every kind of question.
Search engines hand you ten blue links and let you sort it out. Perplexity hands you the answer, with the sources stapled to the back.
Perplexity AI is an answer engine. You ask in natural language, it reads the live web on your behalf, and it returns a synthesized response with numbered citations on every claim. Click a citation, see the source, decide for yourself.
Under the hood, Perplexity routes your question through a frontier large-language model - Sonar, GPT-5, Claude Opus, Gemini, or Grok - and combines it with real-time retrieval. The output reads like a research brief, not a chatbot transcript.
It works on the web, on macOS, Windows, iOS, and Android. There's a WhatsApp bot for one-line questions, a developer API for builders, and Comet - a full AI-native browser - for people who want their tabs to do work for them.
Free to start. $20/month for Pro, with access to every frontier model. $200/month for Max, with unlimited Labs and the agentic Computer feature.
Three years in, shipping faster than ever. The biggest launches of the past few quarters - what landed, when, and why it matters.
Originally a $200/mo Max exclusive, the agentic Computer is now available to all Pro subscribers — 20+ advanced models, prebuilt skills, and hundreds of connectors.
Three months after its $200-only debut, Comet exits the Max gate and becomes free for everyone. Background Assistants and Comet Plus publishers announced in the same drop.
The most powerful AI browser, now in your pocket — built ground-up for mobile, not a port. Cross-tab summarization, voice chat across tabs, ad blocking on by default.
A local-device variant of Computer, announced at Perplexity's first developer conference. Runs on dedicated hardware with persistent file access — sensitive data never leaves your machine.
A new model for personality-driven knowledge: an interactive CR7 experience inside Perplexity, with rare media, curated questions, iconic moments, and the stories behind them.
Deep Research now generates structured presentations and reports natively. Health and Finance verticals get new capabilities in the same release — vertical-specific data sources and templates.
Anything that starts with "find," "explain," or "compare." A few of the most common patterns:
Pull current data on companies, markets, public figures, news, and policy changes - with sources to verify before you cite.
Debug a stack trace, generate boilerplate, explain a library API, or get an opinionated take on architecture trade-offs.
Compare products, services, or destinations side by side on price, specs, reviews, and availability.
Upload a PDF, DOCX, or CSV and ask focused questions. Surface key clauses in contracts, anomalies in data, or themes in long reports.
Hand off a topic and Perplexity reads dozens of sources autonomously, then writes a structured report — citations, gaps, and all.
Generate dashboards, spreadsheets, presentations, and small web apps from a single prompt. Export, share, iterate.
Perplexity routes your question to the model best suited to answer it. You can also pick one explicitly.
Perplexity's in-house model, tuned specifically for grounded web answers with citations. The default for free users.
OpenAI's latest reasoning flagship. Strongest at multi-step logic, math, and structured code generation.
Anthropic's flagship. Excels at long-context reading, careful reasoning, and nuanced writing.
Multimodal — strong on image understanding, table reasoning, and live web tasks. A solid all-rounder.
Real-time, irreverent, plugged into the social pulse. Useful for trends, discourse, and current sentiment.
Video generation orchestrated through Perplexity Labs. Available exclusively on the Max plan.
Every engine Perplexity orchestrates, organized by what it does. House-built and partner alike — pick one explicitly, or let Perplexity route for you.
Default fast answer engine — grounded web search with inline citations on every claim.
Deeper search with more sources per query and richer synthesis. Tuned for substantive research.
OpenAI's frontier general model. Strong at code, structured outputs, and step-by-step explanations.
Anthropic's flagship — best in class for long documents, careful reasoning, and nuanced writing.
Balanced workhorse — fast and capable. Good default when you don't need Opus's depth.
Multimodal all-rounder. Strong on images, charts, tables, and live web understanding.
Real-time, social-aware, irreverent. Plugged into current discourse and trending topics.
Chain-of-thought variant of Sonar. Slows down for multi-step questions and shows its working.
Agentic mode — reads dozens of sources autonomously, then writes a structured report with gaps flagged.
OpenAI's high-effort reasoning engine. Built for problems where careful thought beats fast tokens.
Reasoning variant of GPT-5. More deliberate output, better at code, math, and adversarial logic.
Default image model on Pro. Excellent at photorealistic output and accurate text rendering.
OpenAI's image model. Strong on composition, scene logic, and following complex instructions.
Open-weights image model. Fast, flexible, with broad style coverage from photoreal to illustrative.
Designer-friendly image model with strong layout and graphic-design instincts.
Cinematic short-form video. Best for product visualization, social-ready clips, and storyboards.
Google's video model — strong on physical realism, camera motion, and natural lighting.
Strengths, ideal use cases, and which Perplexity tier each model is available on.
| Model | Strength | Best for | Available on |
|---|---|---|---|
| Sonar | Grounded web answers with citations baked in. | Default search, fact-finding, daily research. | FreeProMax |
| GPT-5.2 | Frontier reasoning & structured output. | Complex coding, analytical write-ups, math. | ProMax |
| Claude Opus 4.6 | Long-context reading & careful writing. | Legal & policy work, drafting, research synthesis. | ProMax |
| Gemini 3 Pro | Multimodal — images, tables, live web. | Visual research, data interpretation. | ProMax |
| Grok 4 | Real-time, conversational, social-aware. | Trends, discourse, current sentiment. | ProMax |
| Deep Research | Agentic — reads dozens of sources autonomously. | Market scans, competitive briefs, dossiers. | ProMax |
| Sora 2 Pro | Cinematic short-form video generation. | Marketing visuals, product demos. | Max |
Three years, nine major releases. From a humble web answer engine in 2022 to a full agentic platform in 2026 — here's how Perplexity got here.
A web-only answer engine debuts at perplexity.ai — built on GPT-3.5 with live web retrieval and numbered citations on every claim. Free for everyone, no signup required.
Perplexity Pro launches at $20/month — initially called "Copilot" — bringing GPT-4 access, follow-up suggestions, file uploads, and Pro Searches that read more sources per query.
Pages turns any research thread into a shareable, formatted article with sections, embeds, and persistent URLs. Spaces (formerly Collections) gain templated workflows.
The Sonar family launches — house-built models specifically tuned for grounded web answers with citations. Sonar Small and Sonar Large open the door to the API and replace third-party models for default search.
Deep Research mode goes live: hand off a topic and Perplexity reads dozens of sources autonomously, then writes a structured report with citations, gaps, and a methodology section.
Native iOS, iPadOS, and Android apps ship with full voice mode, background listening, and Library sync. macOS and Windows desktop apps follow the same month with global hotkeys.
Perplexity Max launches at $200/month with unlimited Labs, priority access to frontier models, and Sora 2 Pro video generation. Comet — Perplexity's AI-native browser — opens to Max users first, with Pro and Free following.
The official Perplexity WhatsApp bot launches at +1 (833) 436-3285 — text, voice notes, and image generation, no signup needed. Spaces gain real-time collaboration with shareable workspaces.
Perplexity Computer launches as a Max-only feature — a cloud-based agent orchestrating 19 different models to handle multi-step workflows autonomously. 10,000 credits per month included; auto-refill optional.
Perplexity adds the latest frontier releases to the Pro and Max model picker the day they ship. Sonar Pro upgraded to a 200k context window. Veo 3.1 joins Sora 2 Pro for video on Max.
They look similar. They are not. ChatGPT is a general-purpose conversational model; Perplexity is an answer engine over the live web. Most heavy users keep both open.
From personal use to enterprise deployment. All plans bill annually with a 16% discount baked in.
For personal, non-commercial use and subject to our policies. For fast answers powered by the latest AI models.
when billed annually
Learn moreKey features:
Designed for teams and organizations streamlining everyday research with accurate answers, securely.
when billed annually
Get startedAll the features of Perplexity Pro, and:
Specifically built for users tackling complex challenges with unlimited queries and the best models.
when billed annually
Get startedAll the features of Enterprise Pro, and:
The complete comparison across Pro, Enterprise Pro, and Enterprise Max - platform, Comet, Computer, security, and support.
| Features | Pro Learn more | Enterprise Pro Get started | Enterprise Max Get started |
|---|---|---|---|
| Platform | |||
| Pro queriesMulti-step reasoning searches and additional sources | Up to 200 per week | 2× Pro queries | 20× Pro queries |
| Deep Research queriesIn-depth research reports on any topic | Up to 20 per month | 2.5× Pro queries | 25× Pro queries |
| Asset generation | Up to 25 per month | 2× Pro assets | 20× Pro assets |
| Video generation | 3 videos per month | 5 videos per month | 15 high-quality videos with audio per month |
| Collaborate in private Spaces with teammates | 5 collaborators per Space | Unlimited collaborators | Unlimited collaborators |
| Ability to choose different models to search on | Select your preferred AI model | Select your preferred AI model | Access to additional advanced AI models and multi-model research mode |
| Get answers from filesFiles must be < 50 MB | 50 uploads per week | 2× Pro uploads | 20× Pro uploads |
| Organizational file repository | — | Configurable controls around file sharing | Configurable controls around file sharing |
| Search through file apps | Attach files from Google Drive and Dropbox | Sync files from Google Drive, SharePoint, and other file apps | Sync files from Google Drive, SharePoint, and other file apps |
| Search and write to apps | — | Get answers and update apps like Salesforce, HubSpot, Slack, and 100+ others | Get answers and update apps like Salesforce, HubSpot, Slack, and 100+ others |
| Memory | — | Configurable memory permissions for admins and users | Configurable memory permissions for admins and users |
| Model council | — | — | Compare your answers with all the top models in one answer |
| Comet | |||
| Comet Assistant | Included | Assistant in Comet | Assistant in Comet, with advanced models powering it |
| Comet Agent | 40 queries per month | 2× Pro queries | 20× Pro queries |
| Computer | |||
| Computer credits† | 500 credits per month | — | 15,000 credits per month |
| Admin Controls + Security | |||
| Single sign on + SCIM* | — | Integrate with your identity provider for SSO + SCIM | Integrate with your identity provider for SSO + SCIM |
| User management | — | Admins can add, remove, and set permissions across team members | Admins can add, remove, and set permissions across team members |
| Insights dashboard* | — | Analyze your usage trends and track adoption | Analyze your usage trends and track adoption |
| Data retention options* | — | Ability to auto-delete files within 1 day | Ability to auto-delete files within 1 day |
| Audit logs* | — | Login attempts, data modifications, configuration changes | Login attempts, data modifications, configuration changes |
| Support | |||
| Support | Direct support on Discord | Dedicated enterprise support | Dedicated enterprise support |
| Data privacy | |||
| Data privacy | Limited data training with the option to opt out | Perplexity and third party LLM partners do not train on your data | Perplexity and third party LLM partners do not train on your data |
| Certified security | |||
| Certified security | SOC 2 Type II | SOC 2 Type II | SOC 2 Type II |
| Cost per month | $20 / month or $200 / year — save 16% | $40 / month per seat or $400 / year — save 16% | $325 / month per seat or $3,250 / year — save 16% |
* Insights dashboard, audit logs, data retention configurability, and SCIM security features are only accessible with 50+ members or 1 Enterprise Max user in your organization.
† Computer credits are available for paying subscribers.
Two flavors. Read the steps if you're in a hurry; watch the videos if you'd rather follow along. Either way, you'll be productive in under ten minutes.
Grab Perplexity for Windows, macOS, iOS, or Android, or just open perplexity.ai in any browser. The Pro features are identical across surfaces.
Use Google, Apple, or any email address. The same account follows you across web, mobile, and desktop, and synchronizes Library and Spaces.
Below the input you can choose a focus: Web, Academic, Reddit/Social, Math, or Writing. Each one filters which index Perplexity reads from.
Tell it the role, the task, the format, and the constraints. Then ask follow-ups in the same thread — Perplexity keeps the context.
Every claim links back to a numbered source. Click through before you cite it elsewhere — the engine is excellent but not infallible.
Hand-picked walkthroughs from the community — pick a topic, follow along, build the muscle memory.
Here's what a typical Perplexity exchange looks like. Notice the citation footnotes — they're the reason serious researchers prefer it.
Compare Perplexity Pro and ChatGPT Plus on price, models, and live web access. Be concise, use bullet points, and cite sources.
Both are $20 per month, but they emphasize different things1,2:
If your work is mostly research and verification, Pro tends to win on time-saved-per-query. If it's mostly drafting and ideation, ChatGPT's persona memory and writing tools have an edge.
The Sonar API exposes Perplexity's grounded answer engine to your application. Send a question, get back a cited, web-aware response.
The endpoint is OpenAI-compatible — same chat/completions shape with messages, system prompts, and streaming. The differentiator is the citations array Perplexity returns alongside the answer.
API usage is metered separately from your consumer plan. A Pro subscription does not include API credits, and API calls are unaffected by Max's "unlimited" web access.
search_domain_filter# curl example — Sonar with citations curl https://api.perplexity.ai/chat/completions \ -H "Authorization: Bearer $PPLX_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "sonar-pro", "messages": [ { "role": "user", "content": "What changed in EU AI Act this quarter?" } ], "return_citations": true, "stream": false }'
Download the native app on every major platform — or try Comet, the AI browser that works for you.
Pick up where you left off. Your Library, Spaces, and threads sync across every device.
By continuing, you agree to Perplexity's Terms of Service and acknowledge our Privacy Policy.