AI Search
Perplexity
Perplexity searches the web and answers with inline citations. It adds Deep Research, model choice, Projects for shared files, and Computer for multi-step tasks. Companies buy Enterprise Pro or Enterprise Max seats.
The 30-Second Take
Pick Perplexity if your teams research the public web every day and will open the sources behind each answer.
It is an AI answer engine with inline citations and models from OpenAI, Anthropic, Google and its own Sonar. Enterprise Pro seats add Projects, connectors, admin controls and a no-training commitment.
- Best fit
- Teams whose work starts with questions about markets, companies, and topics on the public web.
- Main constraint
- A citation can point to a page that does not support its sentence, so findings need a source check.
- First move
- Pilot Enterprise Pro with one research-heavy team and a rule to open cited sources before sharing.
Why
Watch the citations: independent audits put the failure rate between about one claim in seven and one citation in three. Texas State ended its campus trial in June 2026 and moved users to Gemini.
20 practitioners, 89 sources reviewed, Updated Nov 1, 2026How this is measured
- Operating stack
- AI Knowledge Work
Comparisons
How does Perplexity compare with the alternatives?
Perplexity vs ChatGPT
- AI Assistants
- Web Research
Choose Perplexity for cited research and ChatGPT for a general work assistant; ChatGPT Business lists below Enterprise Pro. Perplexity builds every answer around inline citations and offers models from several providers, while ChatGPT runs on OpenAI's models and adds ChatGPT Work and Codex on the same plan. Each sells business seats with web search, file analysis, connectors, admin controls, and an agent for multi-step work.
Perplexity vs Glean
- Enterprise Search
- Company Knowledge Search
Choose Glean for questions about the company and Perplexity for questions about the outside world. Perplexity answers from the public web first and adds the files each user connects or adds to a Project. Glean indexes company apps through 275+ connectors with each source's permissions enforced, and adds web search as an admin-configured tool.
Operator Insights
What are business and product leaders learning from rolling Perplexity out across their teams?
Ethan SmithCEOGraphite
Professional team. Search marketing agency studying AI answer citations, described September 14, 2025.
Lenny's Podcast, published Listen to the episode (opens Lenny's Podcast in a new tab)
Observation
Graphite's study of thousands of questions found Perplexity's citations overlap Google's about twice as much as ChatGPT's.
- Takeaway
- Measure your visibility in Perplexity and ChatGPT separately, with repeated prompts, since they cite different pages.
- What happened
- Graphite tested thousands of questions and found the citation overlap with Google Search results around 35% for ChatGPT and around 70% for Perplexity. Smith concludes the assistants run similar algorithms with very different citations, so a company has to ask each question several times, with variations, to know how often it appears.
“We're seeing that ChatGPT citations and Google Search results are actually not that similar. Perplexity is interestingly more similar to Google than ChatGPT. We did a study looking at thousands of questions and saw the citation overlap with Google Search results was around 35% for ChatGPT and Google, so not that much.”
Ethan Smith, at 36:41 - Podcast episode (Lenny's Podcast)
- Study described by the CEO
Deployment details
Professional team. Search marketing agency studying AI answer citations, described September 14, 2025.
- Podcast episode (Lenny's Podcast)
- Study described by the CEO
- Team
- Graphite's research team
- Workload
- Citation tracking across AI assistants
- Product version
- Perplexity and ChatGPT
- Measurement
- Citation overlap with Google (about 70% for Perplexity, 35% for ChatGPT)
Gabriel CorralEngineer, with three Cloudflare colleaguesCloudflare
Platform operator. Customer sites that blocked PerplexityBot. Report published August 4, 2025.
The Cloudflare Blog, published Read the write-up (opens The Cloudflare Blog in a new tab)
Observation
Cloudflare caught Perplexity using undeclared crawlers on sites that had blocked it, and de-listed it as a verified bot.
- Takeaway
- If your company publishes content, check whether Perplexity can reach pages you block, and ask legal whether that matters before you buy seats.
- What happened
- After complaints from customers who had disallowed Perplexity in robots.txt and blocked its declared crawlers with firewall rules, Cloudflare's tests showed Perplexity reaching those sites through undeclared crawlers with rotating IP addresses and a generic browser user agent. After the incident, on August 4, 2025, Cloudflare de-listed Perplexity as a verified bot and added managed rules that block the stealth crawling for every customer.
“We received complaints from customers who had both disallowed Perplexity crawling activity in their robots.txt files and also created WAF rules to specifically block both of Perplexity’s declared crawlers: PerplexityBot and Perplexity-User.”
Cloudflare - Public engineering write-up
- Tests run by Cloudflare
Deployment details
Platform operator. Customer sites that blocked PerplexityBot. Report published August 4, 2025.
- Public engineering write-up
- Tests run by Cloudflare
- Team
- Cloudflare's bot management engineers
- Workload
- Detecting crawler traffic against robots.txt rules
- Product version
- Perplexity's crawlers
Buying and deployment
What business and product leaders should verify before rolling it out across teams.
Sourced from Perplexity documentation
Signals
What is the current evidence saying about Perplexity?
What people using Perplexity for work and research report publicly, with every finding linked to its sources.
Practitioner sentiment, leaders and teams
−25 Negative net
of 20 sources that take a side, each author counted once. Small sample.
How this figure is madeCurrent sentiment
Positive, mixed and negative sources
Research teams that audited Perplexity's answers report wrong answers and weak sources, while benchmark studies credit its citation accuracy. Browser security teams report that Comet's agent followed hidden instructions and phishing pages. Campus buyers split: Texas A&M approved Enterprise Pro for FERPA data, and Texas State ended its trial.
Run a pilot on questions your team already asks to measure how often the answer is right and cites its source.
Evidence field. All 89 sources, one mark each.
Arrange by
Each mark is one source: positive, mixed, negative, no opinion stated, or documentation (not counted).
| Group | Sources | Positive | Mixed | Negative | No opinion stated and not counted in the score | Documentation, checked and not counted |
|---|---|---|---|---|---|---|
| Positive | 5 | 5 | 0 | 0 | 0 | 0 |
| Mixed | 5 | 0 | 5 | 0 | 0 | 0 |
| Negative | 10 | 0 | 0 | 10 | 0 | 0 |
| Not counted | 5 | 0 | 0 | 0 | 5 | 0 |
| Documentation | 16 | 0 | 0 | 0 | 0 | 16 |
| Outside this group | 48 | 18 | 12 | 16 | 2 | 0 |
| Quarter | Sources |
|---|---|
| Q1 2025 | 15 |
| Q2 2025 | 12 |
| Q3 2025 | 14 |
| Q4 2025 | 5 |
| Q1 2026 | 8 |
| Q2 2026 | 6 |
| Q3 2026 | 13 |
| Q4 2026 | 0 |
| Documentation, checked | 16 |
By sentiment
16 documentation pages check facts and are not counted; 48 sit outside this group.
Hover a mark for a quick look, click it or press Enter to open its source.Tap a mark to open its source, then step to its neighbors with Previous and Next.
What practitioners praise
Each mark is a source on it; filled marks praise it.
Most common concerns
Top concern of 2. Each mark is a source on it; filled marks raise it.
All 89 sources behind these findings
- PricingPerplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- Enterprise Pricing and Billing FAQPerplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- Which Perplexity Subscription Plan is right for you?Perplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- Perplexity MaxPerplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- What advanced AI models are included in my subscription?Perplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- Understanding source labelsPerplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- What is Research mode?Perplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- What is Model Council?Perplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- What are Projects?Perplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- Changelog: Projects (Aug 4, 2026)Perplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- Using File Connectors in PerplexityPerplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- Data Retention for EnterprisePerplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- Are third-party model providers training on my data?Perplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- Comet for EnterprisePerplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- How credits work for Enterprise organizations (admin guide)Perplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- How Credits Work on PerplexityPerplexity (Official Perplexity pricing page, Help Center, and changelog), Product documentation, checked Oct 12, 2026
- Perplexity - Enterprise Pro Frequently Asked QuestionsTexas A&M Technology Services, the university's central IT organization (self-described: Texas A&M University, Enterprise Pro licenses for students, faculty, and staff through June 30, 2027), University IT service page, published Sep 28, 2026
- Comment on: Three sites made 215,128 "best software" pages for AI. Perplexity cites themjwrallie, Hacker News commenter (self-described: Education plan trial), Hacker News comment, published Sep 3, 2026
- Comment on: Three sites made 215,128 "best software" pages for AI. Perplexity cites themGrisu_FTP, Hacker News commenter (self-described: Free Pro year with a banking subscription), Hacker News comment, published Sep 3, 2026
- Comment on: Three sites made 215,128 "best software" pages for AI. Perplexity cites themembedding-shape, Hacker News commenter (self-described: Research use before ChatGPT search), Hacker News comment, published Sep 3, 2026
- Comment on: Three sites made 215,128 "best software" pages for AI. Perplexity cites themAurornis, Hacker News commenter (self-described: Twelve-month free offer), Hacker News comment, published Sep 2, 2026
- Comment on: Three sites made 215,128 "best software" pages for AI. Perplexity cites themstranded22, Hacker News commenter (self-described: Three-year Pro subscriber), Hacker News comment, published Sep 2, 2026
- Comment on: A third of Perplexity's citations don't contain the number they're cited forhek2sch, Hacker News commenter (self-described: Research that needs correct citations), Hacker News comment, published Sep 2, 2026
- Comment on: Three sites made 215,128 "best software" pages for AI. Perplexity cites themsimur, Hacker News commenter (self-described: Free Pro year), Hacker News comment, published Sep 2, 2026
- A third of Perplexity's citations don't contain the number they're cited forHaus Research, an independent research publisher (self-described: Citation audit of Sonar and Sonar Pro on 310 questions about 210 technology companies), Independent benchmark, Haus Research, published Sep 2, 2026
- Three sites made 215,128 "best software" pages for AI. Perplexity cites themTrellner Research, an independent research publisher (self-described: Audit of 7,534 citations from Sonar and Sonar Pro across 380 software categories), Independent benchmark, Trellner, published Sep 2, 2026
- Comment on: Ask HN: Do you feel comfortable admitting that you use AI?NishanStepak, Hacker News commenter (self-described: Looks up news and library articles at work), Hacker News comment, published Aug 18, 2026
- How Perplexity Actually Picks Sources (I Read the Stream, Not the Answers)Suganthan Mohanadasan, co-founder of Snippet Digital (self-described: Solo, AI SEO and content strategy research using a logged-in Pro account), Independent review, published Jul 27, 2026
- Comment on: Microsoft 365 Copilot adoption is under 4.5%, only 1% use it weeklywolvoleo, Hacker News commenter (self-described: Employer allows Microsoft 365 Copilot only), Hacker News comment, published Jul 7, 2026
- Perplexity Said No Business Here: Why I Cancelled Pro and Went Back To FreeTerry Brock, co-host of Stark Raving Entrepreneurs (self-described: Business content creator and speaker using Perplexity for research), Video episode show notes, Stark Raving Entrepreneurs (Substack), published Jun 5, 2026
- Perplexity Free Trial Ended June 1, 2026Texas State University Division of Information Technology, the university's central IT organization (self-described: Texas State University, campus Enterprise Pro trial for students, faculty, and staff), University IT service page, published Jun 1, 2026
- Comment on: Search engines alternatives now that Google isn't Google anymorehamdingers, Hacker News commenter (self-described: General research), Hacker News comment, published May 25, 2026
- Comment on: Google changes its search boxHavoc, Hacker News commenter (self-described: Google Workspace account holder), Hacker News comment, published May 19, 2026
- Comment on: Check your fucking sources, peopleflail, Hacker News commenter (self-described: Research use), Hacker News comment, published May 15, 2026
- LiveResearchBench: A Live Benchmark for User-Centric Deep Research in the WildJiayu Wang and coauthors, University of Wisconsin-Madison and Salesforce (self-described: Benchmark of long-form research reports from search models and deep research agents), Independent benchmark, arXiv, published Apr 18, 2026
- Comment on: Personal Computer by Perplexityhbosch, Hacker News commenter (self-described: Former user), Hacker News comment, published Mar 12, 2026
- "AgenticBlabbering" How AI Browsers' Verbose Reasoning Fuels the Ultimate Scamming MachineShaked Chen, Security Researcher, Guardio (Guardio Labs reading the reasoning stream Comet's agent sends while it browses), Company research blog, Guardio Labs, published Mar 11, 2026
- Perplexity Enterprise Pro trial at Corvinus: 16 March - 17 April 2026Corvinus University of Budapest, university-wide Enterprise Pro trial (self-described: Corvinus University of Budapest, Enterprise Pro trial for its whole community through single sign-on), University news page, published Mar 10, 2026
- Search Arena: Analyzing Search-Augmented LLMsMihran Miroyan, Tsung-Han Wu, and coauthors, University of California, Berkeley (self-described: Analysis of human preference votes on search-augmented models in Search Arena), Research paper, arXiv, published Mar 2, 2026
- Comment on: How will OpenAI compete?codethief, Hacker News commenter (self-described: Used over several years; coworkers report the same), Hacker News comment, published Feb 28, 2026
- Perplexity Computer: What I Built in One Night (Review, Examples, and How It Compares to OpenClaw and Claude)Karo (Product with Attitude), AI product manager and newsletter author (self-described: Solo, building apps and research packets), Newsletter, Product with Attitude (Substack), published Feb 26, 2026
- Comment on: Ask HN: What happens after the AI bubble bursts?paulcole, Hacker News commenter (self-described: Non-programmer scripting tasks at work), Hacker News comment, published Feb 16, 2026
- Business Brain 715: Choosing the right AIDave Hamilton, co-host of Business Brain (self-described: Small-business podcast host's own research), Podcast episode, Business Brain, published Jan 2, 2026
- Comment on: Kimi K2 1T model runs on 2 512GB M3 Ultraspcf, Hacker News commenter (self-described: Pro included with a banking subscription), Hacker News comment, published Dec 15, 2025
- Comment on: Google begins showing ads in AI Mode (AI answers)Agraillo, Hacker News commenter (self-described: Regular user of Perplexity and Google AI Mode), Hacker News comment, published Nov 22, 2025
- Unseeable prompt injections in screenshots: more vulnerabilities in Comet and other AI browsersShivan Kaul Sahib, VP of Privacy & Security, Brave (Brave's privacy and security team extending its agentic browser research to Comet screenshots), Company blog post, Brave, published Oct 21, 2025
- Comet sends all your URLs to Perplexity servers (and there's no way to stop it)Shivan Kaul, privacy engineer and blogger (self-described: Testing Comet's privacy settings), Personal blog post, published Oct 19, 2025
- Is Perplexity Comet Safe? LayerX Finds a Prompt Injection Attack VectorAviad Gispan, Senior Researcher, LayerX (LayerX research team testing Comet with connected email and calendar), Company research blog, LayerX, published Oct 4, 2025
- Lenny's Podcast: Ethan Smith on AI search visibilityEthan Smith, CEO of Graphite (self-described: Search marketing agency's study of AI answer citations), Podcast episode, Lenny's Podcast, published Sep 14, 2025
- LayerX Finds that Perplexity's Comet Browser is Up To 85% More Vulnerable to Phishing and Web Attacks Than ChromePaloma Perlov, Researcher, LayerX (LayerX research team running recent phishing pages against AI and standard browsers), Company research blog, LayerX, published Sep 8, 2025
- How I'm Using PerplexityWilliam, software developer and blogger (self-described: Solo, software and web research), Personal blog post, published Sep 7, 2025
- Agentic Browser Security: Indirect Prompt Injection in Perplexity CometArtem Chaikin, Senior Mobile Security Engineer, Brave (Brave's security team testing Comet while building agentic browsing into its own Leo assistant), Company blog post, Brave, published Aug 20, 2025
- "Scamlexity" We Put Agentic AI Browsers to the Test - They Clicked, They Paid, They FailedNati Tal, Head of Guardio Labs, Guardio (Guardio Labs testing Comet's agent on a fake store and a live phishing page), Company research blog, Guardio Labs, published Aug 20, 2025
- Comment on: Perplexity Makes Longshot US$34.5B Offer for Chromescrumper, Hacker News commenter (self-described: Getting up to speed on new clients and job responsibilities), Hacker News comment, published Aug 12, 2025
- Drop #691: Bye, Bye, Perplexity.aihrbrmstr, author of the Daily Drop newsletter (self-described: Paid Perplexity subscriber for research), Newsletter, published Aug 7, 2025
- Comment on: Perplexity Response to Cloudflareskeledrew, Hacker News commenter (self-described: Regular research use), Hacker News comment, published Aug 5, 2025
- Perplexity is using stealth, undeclared crawlers to evade website no-crawl directivesGabriel Corral and three colleagues, Cloudflare engineers (self-described: Cloudflare tests after customer complaints), Company blog post, Cloudflare, published Aug 4, 2025
- How I Built a Self-Updating AI Sales Agent for Using Perplexity Public RoomsAli Farhat, founder and architect at Scalevise (self-described: Scalevise, sales agent for company services), Company blog post, DEV Community, published Jul 24, 2025
- ResearcherBench: Evaluating Deep AI Research Systems on the Frontiers of Scientific InquiryTianze Xu and coauthors, Shanghai Jiao Tong University (self-described: Academic benchmark of deep research systems on frontier AI research questions), Independent benchmark, arXiv, published Jul 22, 2025
- Comment on: Perplexity launches Comet, an AI-powered web browserfelarof, Hacker News commenter (self-described: Annual Pro subscriber), Hacker News comment, published Jul 10, 2025
- Mind2Web 2: Evaluating Agentic Search with Agent-as-a-JudgeBoyu Gou and coauthors, The Ohio State University (self-described: Academic benchmark of long-horizon search tasks graded by judge agents), Independent benchmark, arXiv, published Jul 3, 2025
- Lenny's Podcast: Andrew Wilkinson's AI stackAndrew Wilkinson, co-founder of Tiny (self-described: Personal meeting-preparation agent), Podcast episode, Lenny's Podcast, published Jul 3, 2025
- The Attribution Crisis in LLM Search ResultsIlan Strauss, Jangho Yang, Tim O'Reilly, Sruly Rosenblat, and Isobel Moure, paper authors (self-described: Analysis of about 14,000 LMArena conversation logs from search-enabled models), Research paper, arXiv, published Jun 27, 2025
- Comment on: Oklo, the Earth's Two-billion-year-old only Known Natural Nuclear Reactor (2018)geocrasher, Hacker News commenter (self-described: Research and discovery), Hacker News comment, published Jun 20, 2025
- DeepResearch Bench: A Comprehensive Benchmark for Deep Research AgentsMingxuan Du and coauthors, University of Science and Technology of China and Metastone Technology (self-described: Academic benchmark of four deep research agents on 100 PhD-level research tasks), Independent benchmark, arXiv, published Jun 13, 2025
- Comment on: Ask HN: Who is still using Perplexity?msephton, Hacker News commenter (self-described: Quick searches on macOS), Hacker News comment, published Jun 9, 2025
- Comment on: Ask HN: Who is still using Perplexity?Rochus, Hacker News commenter (self-described: Research-focused paying user), Hacker News comment, published Jun 9, 2025
- TXST implements AI tools across campus, classesKimberly Conner, IT digital strategies lead at Texas State University (Texas State University, Enterprise Pro for students, faculty, and staff from April 2025), News article, The University Star, published Jun 4, 2025
- Comment on: Ask HN: What are you working on? (May 2025)35mm, Hacker News commenter (self-described: Building a company prospect list), Hacker News comment, published May 26, 2025
- Google Search is losing to PerplexityAshish Bhatia, software developer and blogger (self-described: Solo, Android application debugging), Personal blog post, published May 3, 2025
- Comment on: AI assisted search-based research works nowCSMastermind, Hacker News commenter (self-described: Tested deep research products on a countable fact), Hacker News comment, published Apr 21, 2025
- Comment on: Kagi Assistant is now available to all usersloehnsberg, Hacker News commenter (self-described: Paid for Perplexity and Kagi side by side), Hacker News comment, published Apr 18, 2025
- Comment on: Kagi Assistant is now available to all usersZambyte, Hacker News commenter (self-described: Paid for Perplexity and Kagi for half a year), Hacker News comment, published Apr 18, 2025
- Comment on: Kagi Assistant is now available to all usersmbesto, Hacker News commenter (self-described: Personal project knowledge base), Hacker News comment, published Apr 18, 2025
- [Bug] I'm experiencing a similar issue - truncated JSON in non-streaming mode (stream: false)Piotr Sitkowski, developer on the Sonar API (self-described: PHP app calling Sonar with domain filters for health sources), GitHub issue, published Mar 24, 2025
- Comment on: Claude can now search the webpbronez, Hacker News commenter (self-described: Free year of Pro, compared with Kagi), Hacker News comment, published Mar 21, 2025
- Kogod School of Business Partners with Perplexity to Provide Enterprise AI Access to All Students, Staff and FacultyDavid Marchick, Dean of the Kogod School of Business (self-described: Kogod School of Business, Enterprise Pro for every student, staff member, and faculty member), Organization announcement, published Mar 19, 2025
- [Bug] sonar-pro API does not respect response_formattduffield, developer on the Sonar API (self-described: Pregnancy information app requesting cited answers as JSON), GitHub issue, published Mar 17, 2025
- AI Search Has a Citation ProblemKlaudia Jaźwińska and Aisvarya Chandrasekar, Tow Center for Digital Journalism, Columbia Journalism School (self-described: Tow Center study of eight AI search tools on 1,600 news-excerpt queries), Independent benchmark, Columbia Journalism Review, published Mar 6, 2025
- Comment on: Perplexity Deep Researchlarsiusprime, Hacker News commenter (self-described: Policy research, fifty-state comparison table), Hacker News comment, published Feb 16, 2025
- Comment on: Perplexity Deep ResearchRobinL, Hacker News commenter (self-described: Former regular user), Hacker News comment, published Feb 16, 2025
- Comment on: Perplexity Deep Researchsenko, Hacker News commenter (self-described: Tested on a niche national political topic the commenter knows), Hacker News comment, published Feb 16, 2025
- Comment on: Perplexity Deep Researchvic_nyc, Hacker News commenter (self-described: Uses ChatGPT every day for work), Hacker News comment, published Feb 16, 2025
- Comment on: Perplexity Deep Researchtomaskafka, Hacker News commenter (self-described: Tested on topics the commenter knows), Hacker News comment, published Feb 16, 2025
- Groundbreaking BBC research shows issues with over half the answers from Artificial Intelligence (AI) assistantsPete Archer, Programme Director for Generative AI, BBC (BBC study of four AI assistants answering news questions from BBC News articles), Research announcement, BBC Media Centre, published Feb 11, 2025
- Taking on Google search: Perplexity and KagiGergely Orosz, author of The Pragmatic Engineer newsletter (self-described: Day-to-day web search for newsletter work), Newsletter, published Jan 28, 2025
- Changelog & Friends 76Jerod Santo, co-host of Changelog (self-described: Developer podcast company), Podcast episode, Changelog & Friends, published Jan 17, 2025
- Comment on: Perplexity got adsdtromero, Hacker News commenter (self-described: Baseline research and finding discussion threads), Hacker News comment, published Jan 4, 2025
- Comment on: Perplexity got adsschmorptron, Hacker News commenter (self-described: Free year of Pro), Hacker News comment, published Jan 4, 2025
Jan 2025 to Sep 2026.
73 sources from practitioners, by month and sentiment.
Where It Fits
Where does Perplexity belong in the stack?
Perplexity is Core across Assist and Find and supports Create, Analyze, and Automate.
- Business area
- Company-wide
- Function
- Knowledge Work
- Operating stack
- AI Knowledge Work
- Primary role
- AI Answer Engine & Research Assistant
Assist
AI assistant: Core
AI assistant, core. Every seat gets a general assistant for questions, drafting, and file analysis, with a choice of models on paid plans from providers including OpenAI, Anthropic, Google, and Perplexity itself. Buyers weigh Enterprise Pro against company-wide assistants such as ChatGPT, Claude, and Gemini.
Find
Web research: Core
Web research, core. Web research is Perplexity's primary role, and each answer comes from a web search and carries inline citations. Deep Research reads many sources to write a report, and paid plans add PitchBook, Statista, and other premium sources.
Enterprise search: Supports
Enterprise search, supports. Projects hold shared files and instructions, and file connectors search Google Drive, OneDrive, SharePoint, Dropbox, and Box at query time, with each member connecting their own account. A permission-aware index across all company apps stays with an enterprise search product.
Create
Documents: Supports
Documents, supports. Create files and apps and Computer produce reports, spreadsheets, presentations, and simple web apps from research. The team's documents and wiki stay in its writing tools.
Meeting notes: Touches
Meeting notes, touches. Perplexity's plan pages list no meeting capture. Recordings and notes reach it as uploaded audio, video, or document files, or through connectors such as Slack and Notion, and capture stays with a meeting notes tool.
Analyze
Data analysis: Supports
Data analysis, supports. Deep Research runs calculations in a code sandbox, Create files and apps builds dashboards and spreadsheets, and connectors reach Snowflake and Databricks. Governed reporting on company data stays with an analytics tool.
Automate
Automation: Supports
Automation, supports. Computer runs multi-step tasks on a schedule or when events occur in connected apps, billed in credits. Durable automations across many business systems stay with an automation platform.
Market mapThe layers Perplexity covers, and who else owns each one.
General AI assistant
Web research & cited answers
- No other product we cover owns this layer
Enterprise search & company knowledge
Writing & documents
Data analysis & reporting
2 more layers on the full map. Core, Supports and Touches describe coverage. They say nothing about quality.
What your subscription unlocks
What does signing in with a Perplexity plan give you that a Perplexity API key does not?
IncludedNot on this tierOpens the vendor's own words35 claims from 34 Perplexity pages
Which tier gets what
It performs dozens of searches, reads hundreds of sources, and reasons through the material to autonomously deliver a comprehensive report.
Free users get limited access to Research. For those who need more, Perplexity Pro subscribers get extended access.
Research limits reset on a rolling basis rather than on a fixed calendar date.Perplexity: What is Research mode?2Research modeRuns dozens of searches over hundreds of sources and writes a report; caps reset on a rolling window.LimitFree 1 a month; Enterprise Pro 50 a month; Enterprise Max 500TiersEducation Pro, Perplexity Pro, Enterprise Pro, Perplexity Max, Enterprise Max
Research Queries 1/month Monthly limits (average use) Monthly limits (average use) Monthly limits (advanced use) Extended limits (50/month) Highest limits (500/month)Perplexity: Which Perplexity Subscription Plan is right for you?Free 1 a month; Enterprise Pro 50 a month; Enterprise Max 500Education Pro: includedPerplexity Pro: includedEnterprise Pro: includedPerplexity Max: includedEnterprise Max: included
On web and in the iOS app, Model Council is included for subscribers on the Max and Enterprise Max plans. It is not available on the Free , Pro , Education Pro , or Enterprise Pro plans.
There is a monthly cap on how many Model Council queries you can run
A council on web always runs three modelsPerplexity: What is Model Council?Monthly query cap on web and iOSEducation Pro: not includedPerplexity Pro: not includedEnterprise Pro: not includedPerplexity Max: includedEnterprise Max: included
It is available to Pro and Max subscribers on an NVIDIA DGX Spark, or on a Windows or Linux PC with a supported NVIDIA RTX GPU.
Work handled by the local model does not consume Computer credits
Plan Pro or Max, on a personal or an Enterprise plan
NVIDIA DGX Spark, or an NVIDIA GeForce RTX or RTX PRO GPU with 24 GB of VRAM or morePerplexity: What is Portable Computer?Needs a DGX Spark or an RTX GPU with 24 GB of VRAMEducation Pro: not includedPerplexity Pro: includedEnterprise Pro: includedPerplexity Max: includedEnterprise Max: included
Max Assistant is Comet's next-level mode, powered by cutting-edge reasoning models that intelligently route every request.
As a Max subscriber, you receive the highest weekly limit on browser agent queries available across all subscription tiers.Perplexity: Perplexity MaxEducation Pro: not includedPerplexity Pro: not includedEnterprise Pro: not includedPerplexity Max: includedEnterprise Max: not included
Brain is a self-improving memory system that builds a working model of your projects, people, and files to give Perplexity Computer the context it needs to move your work forward. Brain is rolling out in Research Preview to Max and Enterprise Max subscribers using Computer.Perplexity: What is Brain?Education Pro: not includedPerplexity Pro: not includedEnterprise Pro: not includedPerplexity Max: includedEnterprise Max: included
Show all 5Show fewer
Computer is an independent digital worker that completes tasks and workflows for you.
Computer is available to users with an active subscription.
Computer is available on web desktop, iOS, and Android.Perplexity: What is Computer?5Perplexity ComputerAn autonomous agent that researches, writes documents, builds apps and runs scheduled tasks across connected apps on web and mobile.TiersEducation Pro, Perplexity Pro, Enterprise Pro, Perplexity Max, Enterprise Max
It includes everything in Pro
access to Perplexity ComputerPerplexity: What is Education Pro?Education Pro: includedPerplexity Pro: includedEnterprise Pro: includedPerplexity Max: includedEnterprise Max: included
Browser Agent (Comet Assistant) Queries No Monthly limits (average use) Monthly limits (average use) Monthly limits (advanced use) Extended limits (80/month) Highest limits (800/month)Perplexity: Which Perplexity Subscription Plan is right for you?6Comet browser agentComet Assistant acts on websites inside the Comet browser for desktop and mobile; Free accounts get no agent queries.LimitMonthly caps; Enterprise Pro 80 a month, Enterprise Max 800TiersEducation Pro, Perplexity Pro, Enterprise Pro, Perplexity Max, Enterprise Max
Available for Mac, Windows, iOS, and AndroidPerplexity: Comet Browser: a Personal AI AssistantMonthly caps; Enterprise Pro 80 a month, Enterprise Max 800Education Pro: includedPerplexity Pro: includedEnterprise Pro: includedPerplexity Max: includedEnterprise Max: included
Licensed connectors are available to Perplexity Pro, Max, and Enterprise users who hold an eligible license with the data partner.
FactSet : compliant, finance-grade answers grounded in your authorized FactSet data.Perplexity: Premium Data SourcesEducation Pro: not includedPerplexity Pro: includedEnterprise Pro: includedPerplexity Max: includedEnterprise Max: included
Authenticated integrations with Gmail, Outlook, GitHub, Linear, Slack, Notion, and morePerplexity: What is Computer?11App connectorsPaid plans connect Slack, Gmail, GitHub, Notion and other apps so answers and Computer tasks draw on workspace data.TiersPerplexity Pro, Enterprise Pro, Perplexity Max, Enterprise Max
The Connector for Slack is available to Perplexity Pro, Perplexity Max, Enterprise Pro and Enterprise Max users.Perplexity: Using the Connector for SlackEducation Pro: not includedPerplexity Pro: includedEnterprise Pro: includedPerplexity Max: includedEnterprise Max: included
File and App Creation No Monthly limits (average use) Monthly limits (average use) Monthly limits (advanced use) Extended limits (50/month) Highest limits (500/month)Perplexity: Which Perplexity Subscription Plan is right for you?12Create files and appsBuilds reports, spreadsheets, dashboards and simple web apps from a prompt, under a monthly cap.LimitEnterprise Pro 50 a month; Enterprise Max 500TiersEducation Pro, Perplexity Pro, Enterprise Pro, Perplexity Max, Enterprise Max
Create files and apps has access to the most advanced tools and can craft everything from reports and spreadsheets to dashboards and simple web applications.Perplexity: What is Perplexity Pro?Enterprise Pro 50 a month; Enterprise Max 500Education Pro: includedPerplexity Pro: includedEnterprise Pro: includedPerplexity Max: includedEnterprise Max: included
You send a task to a dedicated address, Computer reads the full email context, runs the task, and replies in the same thread with the result.
A Perplexity plan with Perplexity Computer.
Deliverables such as PDFs, spreadsheets, images, videos, and other artifacts are returned as email attachments.Perplexity: Computer in EmailEducation Pro: includedPerplexity Pro: includedEnterprise Pro: includedPerplexity Max: includedEnterprise Max: included
A Project keeps your searches, Perplexity Computer tasks, files, instructions, and connected tools together
Anyone in [your organization] with access can edit
Enterprise plans only.Perplexity: What are Projects?15Organization-wide ProjectsEnterprise Projects share searches, Computer tasks, files and connectors across the organization; personal accounts share by invite only.TiersEnterprise Pro, Enterprise Max
Each Enterprise Pro seat also includes Perplexity Ask, Projects, and connectors, managed by your administrators.Perplexity: What is Enterprise Pro?Education Pro: not includedPerplexity Pro: not includedEnterprise Pro: includedPerplexity Max: not includedEnterprise Max: included
Enterprise Max users can generate 15 videos per month.Perplexity: Which Perplexity Subscription Plan is right for you?16Veo 3.1 videoGenerates short clips with audio from a prompt or photo; the plan sets the Veo 3.1 variant, and extra video credits are not for sale.LimitUp to 8 seconds each; Enterprise Max 15 a month; allowance refills on a rolling 30 daysTiersPerplexity Pro, Enterprise Pro, Perplexity Max, Enterprise Max
Available now for Max, Pro and Enterprise Pro/Max subscribers on Web app, iOS, and Android.
Each video is up to 8 seconds long and may include audio through Veo 3.1.
At present, there is no option to purchase additional credits; once you've used your monthly allocation, you'll need to wait for credits to replenish on a rolling 30-day basis.Perplexity: Generating Videos with PerplexityUp to 8 seconds each; Enterprise Max 15 a month; allowance refills on a rolling 30 daysEducation Pro: not includedPerplexity Pro: includedEnterprise Pro: includedPerplexity Max: includedEnterprise Max: included
which include GPT Image 1, Nano Banana, and Seedream 4.5
Max and Enterprise Max users get access to Nano Banana Pro
Pro/Enterprise Pro: Unlock access to limited amount of high-quality images, plus additional medium-quality options.Perplexity: Generating Images with PerplexityEducation Pro: not includedPerplexity Pro: includedEnterprise Pro: includedPerplexity Max: includedEnterprise Max: included
What each tier adds
- Education Pro$10 a month for verified students and educatorsIncludes
- Perplexity Computer
- Comet browser agent
- Research mode
- Create files and apps
- and 5 more on this page
- Perplexity Pro$20 a month or $200 a yearEverything in Education Pro, plus
- Veo 3.1 video
- Image generation
- Licensed finance data
- App connectors
- and 4 more on this page
- Enterprise Pro$40 a seat a month or $400 a seat a yearEverything in Perplexity Pro, plus
- Perplexity Max$200 a month or $2,000 a yearEverything in Perplexity Pro, plus
- Enterprise Max$325 a seat a month or $3,250 a seat a yearEverything in Enterprise Pro, plus
Plan login or API key
What signing in with a Perplexity plan gives, beside what a Perplexity API key gives.
Coding tools
With the plan login
- Computer in coding agents4Computer in coding agentsClaude Code or Codex can hand tasks to Perplexity Computer over MCP after an OAuth sign-in to the Perplexity account, with no API key.LimitBilled against the account's Computer creditsTiersEducation Pro, Perplexity Pro, Enterprise Pro, Perplexity Max, Enterprise Max
Computer is available to users with an active subscription.
Perplexity: What is Computer?18Computer in coding agentsClaude Code or Codex can hand tasks to Perplexity Computer over MCP after an OAuth sign-in to the Perplexity account, with no API key.LimitBilled against the account's Computer creditsTiersEducation Pro, Perplexity Pro, Enterprise Pro, Perplexity Max, Enterprise MaxThe Computer MCP Server authenticates users via OAuth 2.0 using your Perplexity account. No API key is required.
Computer usage is billed against your Perplexity account credits.
claude mcp add perplexity-computer --transport http https://www.perplexity.ai/rest/computer/mcp codex mcp add perplexity-computer
Perplexity: Perplexity Computer MCP Server
With an API key
- pplx CLI for agents19pplx CLI for agentsThe pplx command-line tool gives terminals and coding agents web search, and every command runs on an API key.
Install and use Perplexity's pplx command-line interface for web search and query-relevant page snippets from your terminal or coding agent.
Every command needs a Perplexity API key .
Perplexity: Perplexity CLI
Usage and billing
With the plan login
- Monthly Computer credits20Monthly Computer creditsAgent work in Perplexity Computer draws on plan credits; paid Pro instead gets a weekly allowance that resets each week and does not roll over.LimitMax 10,000 credits a month; Enterprise Pro 500; Enterprise Max 15,000; 100 credits equal $1TiersPerplexity Pro, Enterprise Pro, Perplexity Max, Enterprise Max
Computer credits are available to Pro and Max subscribers, including Enterprise Pro and Enterprise Max.
Consumer Max plans start with 10,000 credits a month.
Paid Consumer Pro plans now include a weekly amount of Computer usage.
Your included usage resets every week.
Unused included usage does not roll over from one week to the next.
Today, 100 credits equals $1.
Perplexity: How credits work on Perplexity21Monthly Computer creditsAgent work in Perplexity Computer draws on plan credits; paid Pro instead gets a weekly allowance that resets each week and does not roll over.LimitMax 10,000 credits a month; Enterprise Pro 500; Enterprise Max 15,000; 100 credits equal $1TiersPerplexity Pro, Enterprise Pro, Perplexity Max, Enterprise MaxEach Enterprise Max seat includes 15,000 Perplexity Computer credits every month, compared with 500 on Enterprise Pro.
Perplexity: What is Enterprise Max? - Buy extra credits20Buy extra creditsWhen plan credits run out, subscribers buy more by hand or turn on auto-refill, up to a monthly spending cap.LimitMax cap defaults to $200 a month and can go up to $5,000TiersPerplexity Pro, Perplexity Max
Your plan may include a monthly amount. You can purchase more as needed.
If your credits run out and you have not turned auto-refill on, your tasks will pause until you buy more or your monthly credits reset.
Every Max account has a monthly spending cap on total credit charges per billing period. The default is $200, and you can set it anywhere up to $5,000 in your credit settings. Pro accounts also have a monthly spending cap.
Perplexity: How credits work on Perplexity
Show all 4Show fewer
- Pro Search weekly caps2Pro Search weekly capsPro Search runs automatically on paid plans up to a weekly cap; Free gets three a day.LimitEnterprise Pro 400 a week; Enterprise Max 4,000 a week; consumer plans get unpublished weekly capsTiersEducation Pro, Perplexity Pro, Enterprise Pro, Perplexity Max, Enterprise Max
Pro Searches 3/day Weekly limits (average use) Weekly limits (average use) Weekly limits (advanced use) Extended limits (400/week) Highest limits (4000/week)
Perplexity: Which Perplexity Subscription Plan is right for you?12Pro Search weekly capsPro Search runs automatically on paid plans up to a weekly cap; Free gets three a day.LimitEnterprise Pro 400 a week; Enterprise Max 4,000 a week; consumer plans get unpublished weekly capsTiersEducation Pro, Perplexity Pro, Enterprise Pro, Perplexity Max, Enterprise MaxThere's nothing to turn on: when you use Search , Perplexity applies Pro Search automatically for tasks like multi-step reasoning, detailed research, and complex projects, up to your plan's limit.
Perplexity: What is Perplexity Pro? - File upload allowance2File upload allowancePaid plans raise upload and analysis limits for PDFs, spreadsheets, audio, video and images.LimitPro up to 50 files per project; Enterprise Pro 100 a week; Enterprise Max 1,000 a weekTiersPerplexity Pro, Enterprise Pro, Perplexity Max, Enterprise Max
Increased rate limits for file and attachment uploads and analysis Exclusive support channels Up to 50 file uploads per project
Weekly limits (advanced use) Extended limits (100/week) Highest limits (1000/week)
Perplexity: Which Perplexity Subscription Plan is right for you?12File upload allowancePaid plans raise upload and analysis limits for PDFs, spreadsheets, audio, video and images.LimitPro up to 50 files per project; Enterprise Pro 100 a week; Enterprise Max 1,000 a weekTiersPerplexity Pro, Enterprise Pro, Perplexity Max, Enterprise MaxEffortlessly upload files
including PDFs, CSVs, audio, video and images
and explore their contents
Perplexity: What is Perplexity Pro?
With an API key
- Prepaid API credits22Prepaid API creditsThe API bills from prepaid credits bought in the API Console, kept apart from Computer credits, with optional automatic top-up.LimitPurchased credits never expire; some new accounts get a one-time $10 credit that lapses after 60 days
API credits are available for purchase on a pay-as-you-go basis.
You can also enable "Automatic Top Up" to automatically replenish your balance when it drops below $2.
Credits you purchase do not expire.
Some new accounts receive a one-time $10 signup credit. The signup credit expires 60 days after it is added
Perplexity: API Payment and billing23Prepaid API creditsThe API bills from prepaid credits bought in the API Console, kept apart from Computer credits, with optional automatic top-up.LimitPurchased credits never expire; some new accounts get a one-time $10 credit that lapses after 60 daysNote: API Credits are separate from Computer Credits .
Perplexity: What is the Perplexity API Platform? - Spend-based rate tiers25Spend-based rate tiersRate limits rise automatically with lifetime API spend, and a reached tier never drops.LimitAgent API from 1 request a second at Tier 0 to 33 at Tier 4 and above
Usage tiers determine your rate limits and access to beta features based on your cumulative API spending.
once you reach a tier, you keep it permanently with no downgrade.
Tier 0 1 QPS Tier 1 3 QPS Tier 2 8 QPS Tier 3 17 QPS Tier 4 33 QPS Tier 5 33 QPS
Perplexity: Rate Limits & Usage Tiers
Show all 4Show fewer
- No API credit in plans2No API credit in plansEnterprise seats carry no API usage and the API needs no subscription, so API spend is a separate purchase on every tier.
No complimentary API credits
Perplexity: Which Perplexity Subscription Plan is right for you?22No API credit in plansEnterprise seats carry no API usage and the API needs no subscription, so API spend is a separate purchase on every tier.No, a Perplexity subscription is not required to purchase API credits or use the API.
Perplexity: API Payment and billing24No API credit in plansEnterprise seats carry no API usage and the API needs no subscription, so API spend is a separate purchase on every tier.No, API usage is not included with Enterprise Pro or Enterprise Max. Users who need API access must purchase credits or top off their balance with a valid payment method.
Perplexity: Enterprise Pricing and Billing: Frequently Asked Questions - No app features on a key2No app features on a keyAn API account opens no advanced models or Pro features in the Perplexity web or mobile app.
No access to advanced models or Pro features within the web/app interface
Perplexity: Which Perplexity Subscription Plan is right for you?
Models
With the plan login
- Third-party model picker5Third-party model pickerPaid plans pick third-party models such as GPT-6.1 Sol or Grok 4.7 by hand inside Search; Free gets only the automatic choice.TiersEducation Pro, Perplexity Pro, Enterprise Pro, Perplexity Max, Enterprise Max
It includes everything in Pro
access to Perplexity Computer, access to Learn Mode , unlimited Pro Searches, file and image uploads, access to premium models
Perplexity: What is Education Pro?26Third-party model pickerPaid plans pick third-party models such as GPT-6.1 Sol or Grok 4.7 by hand inside Search; Free gets only the automatic choice.TiersEducation Pro, Perplexity Pro, Enterprise Pro, Perplexity Max, Enterprise MaxPaid plans can select models as shown below, subject to usage limits.
On the Free plan you can use Best and a limited number of Pro Searches, but you cannot choose an advanced Search model manually.
Pro and Max plans Model Provider Pro Max GPT-6.1 Sol OpenAI Yes Yes
Grok 4.7 xAI Yes Yes
Perplexity: What advanced AI models are included in my subscription? - Top models above Pro8Top models above ProMax and both Enterprise tiers add GPT-6 Astra, Claude Opus 5.5, Claude Fable 5.1 and Gemini 4 Argon to the Search picker; Pro lacks all four.LimitInside the Perplexity apps only; API calls bill separatelyTiersEnterprise Pro, Perplexity Max, Enterprise Max
This only applies within Perplexity web and does not apply to programmatic access via API , which is billed separately.
Perplexity: Perplexity Max26Top models above ProMax and both Enterprise tiers add GPT-6 Astra, Claude Opus 5.5, Claude Fable 5.1 and Gemini 4 Argon to the Search picker; Pro lacks all four.LimitInside the Perplexity apps only; API calls bill separatelyTiersEnterprise Pro, Perplexity Max, Enterprise MaxGPT-6 Astra OpenAI No Yes Gemini 4 Argon Google No Yes
Claude Opus 5.5 Anthropic No Yes Claude Fable 5.1 Anthropic No Yes
Enterprise Pro and Enterprise Max plans Model Provider Enterprise Pro Enterprise Max GPT-6.1 Sol OpenAI Yes Yes GPT-6 Astra OpenAI Yes Yes Gemini 4 Argon Google Yes Yes
Claude Opus 5.5 Anthropic Yes Yes Claude Fable 5.1 Anthropic Yes Yes
Perplexity: What advanced AI models are included in my subscription?
With an API key
- Agent API model catalog27Agent API model catalogThe Agent API calls third-party models from OpenAI, Anthropic, Google and xAI at per-token rates, with built-in tools billed per call.LimitWeb search tool $2.50 per 1,000 calls
access third-party models from OpenAI, Anthropic, Google, and xAI with web search tools and effort levels
Perplexity: Overview28Agent API model catalogThe Agent API calls third-party models from OpenAI, Anthropic, Google and xAI at per-token rates, with built-in tools billed per call.LimitWeb search tool $2.50 per 1,000 callsAll charges are based on actual token consumption, and every API response includes exact token counts so you know your costs per request.
Perplexity: Agent API Models29Agent API model catalogThe Agent API calls third-party models from OpenAI, Anthropic, Google and xAI at per-token rates, with built-in tools billed per call.LimitWeb search tool $2.50 per 1,000 callsweb_search (standard) $0.0025 per invocation Standard web search with search_type: "web" . $2.50 per 1,000 invocations
Perplexity: Pricing - Embeddings and Decisions APIs27Embeddings and Decisions APIsAPI-only pplx-embed models serve semantic search and RAG, and the Decisions API returns probabilities for yes/no and scored questions.LimitEmbeddings from $0.004 per million tokens; Decisions from $0.01 per million input tokens
Embeddings API Generate high-quality embeddings for semantic search and RAG pipelines. Decisions API Get typed answers with probabilities for yes/no, multiple-choice, and rubric questions about text and images.
Perplexity: Overview29Embeddings and Decisions APIsAPI-only pplx-embed models serve semantic search and RAG, and the Decisions API returns probabilities for yes/no and scored questions.LimitEmbeddings from $0.004 per million tokens; Decisions from $0.01 per million input tokenspplx-embed-v1-0.6b 1024 $0.004
pplx-decider-flash-v1.2 $0.01 Free None
Perplexity: Pricing
Show all 3Show fewer
- Sonar folds into Agent API30Sonar folds into Agent APISonar Chat Completions support ended September 27, 2026; synchronous calls are being rerouted to the Agent API and async Sonar calls no longer work.
Sonar Chat Completions support ended on September 27, 2026. Synchronous and streaming requests keep working: they are being reformulated as Agent API requests, rolling out gradually by model. Asynchronous Sonar requests are not reformulated and are no longer supported
Perplexity: Migrate from Sonar to the Agent API
Building on it
With the plan login
No difference statedWith an API key
- Search API31Search APIOne endpoint returns ranked web results as JSON for your own code or model, billed per request with no token charges.Limit$5 per 1,000 requests, $1 with Fast Search; 50 query units a second on every tier
You send a query and get back ranked web results as structured data for your own code to filter, store, or hand to a model. There is no generated answer and no token billing.
Web search $5.00 per 1,000 successful requests Fast Search ( search_type: "fast" ) $1.00 per 1,000 successful requests
Rate limit 50 query units per second, same for every usage tier.
Perplexity: Search API Overview - Batch API32Batch APILarge JSONL jobs run asynchronously outside the request rate limit at normal prices; zero-retention organizations cannot use it.LimitCompletion target 24 hours, hard deadline 48 hours; inputs and outputs kept 30 days
The Batch API runs many Agent API requests asynchronously.
Requests that run inside a batch do not count toward your Agent API request rate limit .
The Batch API is available to every API organization, except organizations with zero data retention. Batch keeps inputs and outputs for 30 days.
There is no separate batch price.
get completion within 24 hours, the completion_window
24h is the only value and the default. It is a target, not a deadline.
The batch's expires_at is its deadline, 48 hours after created_at .
Perplexity: Batch API
Show all 3Show fewer
- No fine-tuning33No fine-tuningThe API offers no fine-tuning of any model.
Do You Support Fine-Tuning? Currently, we do not support fine-tuning.
Perplexity: Frequently Asked Questions
Data and admin
With the plan login
- Training controls2Training controlsEnterprise data is never logged or used for training; consumer paid plans can opt out of data collection in settings.TiersEducation Pro, Perplexity Pro, Enterprise Pro, Perplexity Max, Enterprise Max
Perplexity Pro, Education Pro and Max: You can opt out of data collection in your settings. Enterprise Pro and Max: Your data is never logged or used for training. Admins can enforce strict privacy and organization-level controls.
Perplexity: Which Perplexity Subscription Plan is right for you? - SCIM and audit logs21SCIM and audit logsOne Enterprise Max seat turns on SCIM, audit logs, custom retention and usage insights for the whole organization; without one, it takes 50 Enterprise Pro seats.LimitSession files deleted after seven days on every Enterprise planTiersEnterprise Pro, Enterprise Max
Organization-wide access to SCIM , Audit Logs , configurable Data Retention , and the Insights dashboard for organizations of any size, as long as you have at least one Enterprise Max seat. Without an Enterprise Max seat, these features need a minimum of 50 Enterprise Pro seats.
Files attached to sessions are deleted after seven days on every Enterprise plan.
Perplexity: What is Enterprise Max?
With an API key
- No training, ZDR by contract34No training, ZDR by contractAPI content is never used to train models; zero data retention needs a written agreement and works only with supporting models.
No training by default: Under the API Terms of Service, Perplexity does not use Customer Content to train generative AI models.
ZDR is available only to eligible customers through an account-level configuration under a written agreement with Perplexity.
Not every model supports ZDR.
Perplexity: Privacy & Security

