Last updated: 2026-10-22
What changed
Perplexity AI is an answer engine that searches the live web for each question and attaches numbered citations to the reply. Paid subscribers can choose between models from six AI labs for each query, and the agent product Perplexity Computer runs multi-step tasks on a monthly allowance of credits.
About Perplexity AI
Perplexity AI is an answer engine from Perplexity AI. You ask a question, it searches the live web, and it replies with an answer that carries numbered citations you can click to check. That sourcing is why people choose it over a general chatbot when a claim has to be verified: a student checking a statistic, an analyst pulling recent figures, a journalist confirming a quote. It runs on the web and in iOS, Android, Mac and Windows apps, and since 2026 it has grown past search into an agent product, Perplexity Computer, which works through multi-step tasks using credits that the paid plans include.
Perplexity does not depend on a single model. The Pro page of the vendor lists models from OpenAI, Google, Anthropic, Moonshot AI, Z.ai and xAI, among them GPT-6.1 Sol, Gemini 3.8 Flash, Claude Sonnet 5.5, Kimi K3, GLM 5.3 and Grok 4.7, and Pro and Max subscribers can choose a preferred model for each question. Model Council, which runs one question across several models so you can compare the answers, is part of the paid plans. Paid plans also add premium data sources, with PitchBook, Statista and S&P Capital IQ named on the vendor's pages. Comet is the company's AI browser and has its own HokAI page.
The vendor's pricing page offers a free plan with limited weekly usage, individual Pro and Max plans, and per-seat Enterprise Pro and Enterprise Max plans for teams. Organisations that need role-based access control, a custom domain or volume discounts are sent to sales for a custom quote. The page quotes the rate for annual billing and does not display a month-to-month price, so confirm that at checkout before assuming one. Both individual paid plans currently advertise limited-time Computer credit bonuses. Exact figures live in the pricing notes and the cost question below.
Who should look elsewhere: if most of your work is drafting long documents, writing code or reasoning through a problem across a long conversation, ChatGPT is the better home for that, and so is Claude, because Perplexity is built around the single sourced question rather than the long working session. Reviewers split sharply too: HokAI's stored ratings show G2 at 4.7 from 47 reviews and Trustpilot at 1.6 from 180, with billing and cancellation the recurring complaint, so read the terms before taking an annual plan. If your main need is Chinese-language work, compare Kimi, and for a Google-ecosystem alternative see Gemini. Browse the wider AI search and discovery category before you commit.
Pricing
Prices checked on the vendor's pricing pages on 22 October 2026. Free is $0 with limited weekly usage. Pro is $17 a month when billed annually and Max is $167 a month when billed annually; no month-to-month price is shown on the page.
Enterprise Pro is $34 per seat per month and Enterprise Max is $271 per seat per month, both billed annually, with custom pricing above that for role-based access control and volume discounts. Computer credits are included per plan; see Perplexity Comet for the browser and ChatGPT for a general-purpose comparison. API pricing sits in the separate developer docs and was not re-checked in this pass.
| Tier | Monthly price | What it includes |
|---|---|---|
| Free | Free | Limited weekly usage, citations in every answer, basic AI models |
| Pro | $17/mo | Billed annually. |
| Enterprise Pro | $34/mo | Per seat, billed annually. |
| Max | $167/mo | Billed annually. |
| Enterprise Max | $271/mo | Per seat, billed annually. |
Feature Comparison by Tier
| Feature | Max | Pro | Free | Enterprise Max | Enterprise Pro |
|---|---|---|---|---|---|
| Live web search with citations | ✓ | ✓ | ✓ | ✓ | ✓ |
| Preferred model selection | ✓ | ✓ | — | ✓ | ✓ |
| Monthly Computer credits | 10,000 | — | — | 15,000 | 500 |
| No training on your data | — | — | — | ✓ | ✓ |
| SSO | — | — | — | ✓ | ✓ |
| SCIM, audit logs, retention settings | — | — | — | ✓ | Contract |
Key Features
- Cited Answers from the Live Web: Every answer carries numbered inline citations that link to the source page, and the paid plans add premium data sources such as PitchBook, Statista and S&P Capital IQ.
- Perplexity Computer: An agent that works through multi-step tasks using credits: Max includes 10,000 monthly credits and Enterprise Max 15,000 per seat, against 500 per seat on Enterprise Pro.
- Model Choice: Pro and Max can pick a preferred model per question from a list that includes GPT-6.1 Sol, Claude Sonnet 5.5, Gemini 3.8 Flash, Kimi K3, GLM 5.3 and Grok 4.7.
- Model Council: Compares answers from several models on a single question, and is listed among the paid-plan features.
- Deep Research: Pro adds deep research and Max raises the limits with expert-level research for long investigations and large files.
- Enterprise Security Controls: Enterprise plans add SSO, with SCIM plus audit logs on Enterprise Max, and the vendor publishes its compliance attestations alongside the plans.
Pros
- Numbered citations on every answer make a claim checkable in one click, which is the main reason to pick it over a general chatbot.
- Paid plans let you re-run a question on a different model without leaving the product, which helps when one answer looks off.
- Enterprise buyers get a written commitment that customer data is not used for training, plus a published compliance list.
Cons
- The free plan is limited to weekly usage, and Computer and deep research need a paid plan.
- Trustpilot averages 1.6 of 5 across 180 reviews, mostly about billing and cancelling, so annual plans deserve a careful read of the terms.
- The pricing page shows annual-billed rates only, so the month-to-month cost has to be confirmed at checkout.
- Long-form writing and code help are weaker than in ChatGPT or Claude.
Product Information
- Cloud
- Yes
- Self-Hosted
- No
- On-Premise
- No
- Languages
- English, Chinese, Spanish, French, German, Japanese, Portuguese
- Training
- Documentation, Tutorial videos, Blog articles, Community Discord, Webinars
Data Handling
- Training-data policy
- Enterprise Pro and Enterprise Max do not use customer data to train models (vendor pricing page). Individual-plan data settings were not re-verified on 22 October 2026.
- Data retention
- 30 days
- Compliance
- SOC 2 Type II · HIPAA · GDPR · PCI DSS (Enterprise plans)
Frequently Asked Questions
What does Perplexity cost in 2026?
Subscribers pay $17 a month for Pro or $167 for Max when billed annually, while teams pay $34 or $271 per seat on the two Enterprise plans on the same annual basis. The free tier costs nothing. Figures come from the vendor pricing pages as of 22 October 2026.
Is the free plan enough for everyday searching?
It covers occasional questions: you get cited answers from basic models, but usage is limited each week. Perplexity Computer, deep research and model selection sit on the paid plans, so heavier research outgrows it quickly.
Does Perplexity train its models on my data?
On the two Enterprise plans the vendor states that customer data is not used to train models. Data settings for individual accounts were not re-verified in this pass, so check the privacy controls in your account before pasting in anything sensitive.
Should you research with Perplexity or ChatGPT?
Use Perplexity when the answer must be sourced and checkable, since each reply links its citations. Use ChatGPT when the work is drafting, coding or a long back-and-forth, where a conversation matters more than a citation trail.
What do Pro and Max add over the free plan?
Pro adds deep research, model selection, polished documents and apps, and expanded Computer access with bonus credits. Max adds maximum Computer usage with 10,000 monthly credits, expert-level research, frontier models and the highest limits.
Top Alternatives
- ChatGPT: General conversation, writing help, and coding lean toward ChatGPT; a question that needs a sourced, verifiable answer instead of a generated one leans toward Perplexity.
- Claude: Claude is built for longer, more careful reasoning and writing across a conversation; Perplexity is built for a single question that needs current information with the sources linked.
- Kimi: Kimi carries an edge for Chinese-language tasks specifically; Perplexity is the broader pick for English-language research backed by real-time web access.
HokAI guides covering Perplexity AI
- Best AI Search Engines in 2026: 6 Free Alternatives to Google: Perplexity, Bing Copilot, Komo AI, Andi Search, Consensus and Google Lens compared on free-tier limits, pricing and data handling for switching off Google.
- Best AI Assistant Apps for Android in 2026, Now That Google Assistant Is Dead: Google Assistant shut down on Android September 4, 2026. Here is how Gemini, ChatGPT, Perplexity, Copilot, Meta AI and Claude compare as its replacement.
- Best AI Search Tools in 2026: Match the Tool to the Job: Perplexity, Bing, and Glean all get called AI search tools, but only Glean searches a company's internal Slack. Here is how the category splits by job in 2026.
- ChatGPT Can Still Shop For You. It Just Can't Check You Out Anymore.: ChatGPT still researches best, but checkout moved to merchants. Compare it with Google Universal Cart and Perplexity Instant Buy, re-checked October 2026.
- CiteLens vs Palmata: Which AEO Tool Should You Use in 2026?: CiteLens tracks four AI engines from $249/month; Palmata's Sounder Discovery Agent explains why from $600, now backed by Salesforce. Full pricing and verdict.
- Best AI for Deep Research in 2026: What Each One Can Actually See: ChatGPT, Gemini, Perplexity and Grok each ship Deep Research, but they read different data and cap runs differently. Here is how to pick the right one.