Perplexity
Perplexity is a cited answer engine that combines real-time web search with advanced language models, but its $20 Pro tier is licensed for personal non-commercial use only, its Pro Search allowance is an unpublished “weekly limit,” and its API is a separate pay-as-you-go billing system.
What is Perplexity?
Perplexity is an answer engine, not a general-purpose assistant in the ChatGPT sense. Its primary output is a synthesized answer with inline citations to live web sources, produced by a retrieval pipeline that searches the web, selects relevant passages, and then passes them to a language model for composition. The product is designed around research, fact-finding, and source-backed verification rather than open-ended conversation, code generation, or creative writing.
The platform has expanded beyond the core answer interface. Comet is a Chromium-based AI browser that integrates Perplexity’s assistant into browsing, with the ability to summarize tabs, compare products across sites, draft emails, and automate form-filling under user supervision. The API Platform exposes Sonar models for developers who need grounded answers in their own applications. Model access spans providers: Pro and Max users can select from advanced models offered by OpenAI, Anthropic, Google, xAI, and others, though the selected model may be substituted during periods of heavy usage.
Perplexity’s commercial structure has a critical boundary that is easy to overlook. The consumer Terms of Service license Free, Pro, and Max for personal, non-commercial use only. A $20/month Pro subscription does not authorize client work, revenue-generating research, or internal business workflows. Organizations must purchase Enterprise Pro or Enterprise Max seats, which are priced at $40 and $325 per seat per month respectively. Enterprise seats also change the data-training posture: Enterprise data is never used for model training, while Pro and Max require the user to opt out.
How does Perplexity work?
A user submits a query in natural language. Perplexity classifies the query and routes it to a search-and-synthesis pipeline. Basic searches return a cited answer with relatively shallow retrieval. Pro Search performs a deeper, multi-step retrieval, reformulating the query, gathering more sources, and producing a more thorough answer. Deep Research extends this further, running an autonomous multi-source investigation that can take minutes and produce a longer report.
The selected model affects the reasoning and writing layer, not the retrieval layer. A user on Pro can choose among advanced models, but the underlying web search and citation selection remain Perplexity’s. When a selected model is unavailable—often during heavy usage weeks—Perplexity substitutes another model without necessarily flagging the switch at the point of answer. This is documented in the help centre as a fallback mechanism, not an error state.
Computer adds an agentic execution layer. It accepts a goal, gathers context from files and connectors, and carries out multi-step work across the web and the user’s files. It is metered by credits: 100 credits equals $1, and task costs range from 15–70 credits for a quick fact to 15,000+ credits for an entire app build. Pro does not include a recurring monthly credit allocation; it includes weekly Computer usage and a one-time 4,000-credit bonus. Max includes 10,000 monthly credits.
The API Platform is a separate product with its own billing. Sonar is the lightweight search model, priced at $1 per million input and output tokens with a per-request fee of $5–$12 depending on search context size. Sonar Pro raises token rates to $3 input / $15 output and the request fee to $14 per 1K. Sonar Deep Research adds citation tokens, reasoning tokens, and search query fees. These rates are not interchangeable with web-plan subscriptions.
Use Cases
- Cited research on current events, market developments, or factual questions where source attribution matters
- Using Comet to compare products across multiple shopping sites and summarize reviews in a single browser workspace
- Running a Deep Research report on a complex topic within the Pro plan’s monthly Research allowance
- Selecting advanced models for reasoning-heavy queries, accepting that model access may be substituted during peak load
- Building a cited answer pipeline through the Sonar API for an application, billed separately from any web subscription
- Deploying Enterprise Pro or Enterprise Max where commercial use, data non-training, and seat management are required
Pricing
Pricing model: Freemium
| Plan | Price | Billing | Included usage | Notes |
|---|---|---|---|---|
| Free (Standard) | $0 | free | Practically unlimited basic searches. 3 Pro Searches per day. 1 Research query per month. No access to advanced AI models, image generation, or file and app creation. File uploads limited. Comet Assistant queries not included. Data privacy standard, with no opt-out stated on the plan-comparison table. | — |
| Pro | $20 | per month | $200/year available. Pro Searches subject to “weekly limits (average use)” rather than a published number. Research queries subject to “monthly limits (average use).” Comet Assistant subject to “monthly limits (average use).” Access to advanced models, image generation, and file/app creation with monthly limits. File uploads subject to weekly limits. One-time new-subscriber bonus of 4,000 Computer credits (expire in about 30 days); no recurring monthly credit allocation, but weekly Computer usage is included. Opt-out of AI training available in settings. | — |
| Max | $200 | per month | Everything in Pro, plus Pro Searches subject to “weekly limits (advanced use),” Research and Comet Assistant subject to “monthly limits (advanced use),” highest level of access to advanced models, 10,000 Computer credits per month, one-time new-subscriber bonus of 35,000 credits (expire in about 30 days), early access to new products, and priority support. Opt-out of AI training available. | — |
| Enterprise Pro | $40 | per seat/month | $400/year per seat (approximately 16.7% discount versus monthly). 400 Pro Searches per week, 50 Research queries per month, 80 Comet Assistant queries per month, 100 file uploads per week. 500 Computer credits per month, 8,000 one-time bonus credits. Data never used for model training. Seat management, centralized billing, organization-level file controls, and dedicated support. | — |
| Enterprise Max | $325 | per seat/month | $3,250/year per seat. 4,000 Pro Searches per week, 500 Research queries per month, 800 Comet Assistant queries per month, 1,000 file uploads per week. 15,000 Computer credits per month, 40,000 one-time bonus credits. Highest access to advanced models, extended Create files and apps limits, enhanced video generation (15 eight-second videos per month), SCIM, audit logs, configurable data retention, and priority support. | — |
| API Platform | Pay-as-you-go | usage-based | Separate billing system from web plans. Sonar: $1 per 1M input and output tokens, $5–$12 per 1K requests depending on search context size. Sonar Pro: $3 per 1M input, $15 per 1M output, $14 per 1K requests. Sonar Deep Research: $2 per 1M input, $8 per 1M output, $2 per 1M citation tokens, $5 per 1K search queries, $3 per 1M reasoning tokens. Agent API tools billed separately (e.g., web_search $0.0025 per standard invocation). No access to advanced models or Pro features in the web interface. | — |
Common Limitations
- Consumer tiers are licensed for personal, non-commercial use only · Perplexity’s consumer Terms of Service restrict use of the Services to “personal, non-commercial use only” and explicitly prohibit “exploiting the Services for any commercial purpose.” This applies to Free, Pro, and Max. A $20/month Pro subscription does not grant commercial rights. Businesses or individuals using Perplexity for client work, revenue-generating research, or internal company workflows must purchase Enterprise Pro ($40/seat/month) or Enterprise Max ($325/seat/month). This is the single most expensive and most frequently missed boundary in Perplexity’s pricing structure: the consumer Pro plan is priced like a professional tool but licensed like a personal one.
- Pro Search limits are described as “weekly limits (average use)” without a published number · Perplexity’s official plan-comparison table lists Pro Search on Pro as “Weekly limits (average use)” rather than a specific count. The same table gives Enterprise Pro a concrete figure of 400 per week and Enterprise Max 4,000 per week, but consumer Pro is left opaque. Third-party analyses of code and user reports place the effective range between roughly 200 and 300 per week in some sources, and 100 to 150 per week in others. Perplexity’s support documentation has acknowledged that exact numbers are not disclosed. Users cannot determine in advance how many Pro Searches a $20/month plan includes, and the limit may shift with system load or model availability.
- Advanced model access can degrade during heavy usage periods · Perplexity’s help centre states that on Pro, “access to advanced models may be limited during weeks of especially heavy usage,” and that when a selected model is unavailable, “Perplexity uses another model until access recovers.” This means that a Pro subscriber who selected a specific model—GPT-5 Thinking, Claude Opus 4.6, or another listed model—may receive a response from a different model without explicit confirmation at the point of answer. The selected model is not guaranteed on every request.
- API Platform is a separate billing system with no shared credits · The API Platform is not included in Pro, Max, or Enterprise web plans. It is a pay-as-you-go system with its own credit purchase and rate card. API usage does not draw from and does not replenish web-plan allowances, and web-plan subscriptions do not include complimentary API credits. A developer who needs both the web interface and programmatic access pays into two separate pools. API pricing is token-based and includes per-request fees for Sonar models, separate from the flat monthly subscription price.
- Computer credit consumption is highly variable and Pro has no recurring monthly allocation · Perplexity Computer uses a credit system where 100 credits equals $1. A quick fact or definition costs roughly 15–70 credits; a multi-step research report costs 350–950 credits; a data dashboard costs 2,400–9,800 credits; and an entire app or multi-day build costs 15,000 or more. Pro does not include a recurring monthly credit allocation. It includes weekly Computer usage before credits are drawn, plus a one-time 4,000-credit bonus that expires in about 30 days. Once the weekly included usage and one-time bonus are exhausted, Computer work requires purchasing additional credits. Max includes 10,000 monthly credits, which covers roughly 10 multi-step research reports or one substantial data dashboard, not an unlimited workload.
- Data training opt-out is available on Pro and Max but is not the default in the same way as Enterprise · Perplexity’s plan-comparison table lists Pro and Max data privacy as “Opt-out AI training available,” while Enterprise Pro and Enterprise Max are listed as “Data never used for model training.” This means that on a consumer Pro or Max plan, data may be used for training unless the user actively changes a setting. Enterprise customers receive non-training as a default contractual term. Users handling confidential material on a consumer plan must locate and enable the opt-out rather than relying on a default exclusion.
Key Features
- Real-time web search with inline citations in every answer
- Pro Search for deeper multi-step retrieval and synthesis
- Deep Research for autonomous multi-source reports
- Comet AI browser with tab summarization, shopping comparison, and supervised task automation
- Model selection across providers (OpenAI, Anthropic, Google, xAI, and others) on paid plans
Perplexity vs. ChatGPT and Google Gemini
This comparison focuses on documented differences in primary product orientation, commercial licensing, usage-limit transparency, and billing structure. It does not rank answer accuracy or source quality, which depend on query type, retrieval conditions, and model version.
| Tool | Positioning | Key differences | Best fit | Pricing note |
|---|---|---|---|---|
| Perplexity Subscription plan comparison How credits work (Computer) API pricing | A cited answer engine built around real-time web retrieval, with a separate AI browser (Comet) and a token-priced Sonar API. | Optimized for source-backed answers rather than open-ended generation. Consumer tiers (Free, Pro, Max) are licensed for personal, non-commercial use only; commercial use requires Enterprise Pro or Enterprise Max. Pro Search limits on consumer Pro are described as “weekly limits (average use)” without a published number. API is a separate pay-as-you-go system. Advanced model access may be substituted during heavy usage. Data training is opt-out on Pro/Max, never-on for Enterprise. | Research and fact-finding where inline citations and live web retrieval matter, and where the user is either a personal user or an organization prepared to purchase Enterprise seats. | Free $0 (3 Pro Searches/day, no advanced models). Pro $20/mo or $200/yr (personal use only, unpublished weekly Pro Search limits). Max $200/mo (10,000 Computer credits, advanced weekly limits). Enterprise Pro $40/seat/mo (400 Pro Searches/week, data never trained). Enterprise Max $325/seat/mo (4,000 Pro Searches/week). API: token-based, from $1/1M tokens plus per-request fees. |
| ChatGPT ChatGPT Plus plan ChatGPT pricing | A general-purpose AI assistant for writing, coding, analysis, image generation, and research, with a broader feature surface than a search-first product. | ChatGPT Plus at $20/month is positioned as a professional and productivity tool, with no personal-use-only restriction in its consumer tier. It includes GPT-6 reasoning models, image generation, Codex, Deep Research, Projects, scheduled tasks, and Custom GPTs. The product is optimized for multi-turn assistance and task completion, not solely cited retrieval. Perplexity’s differentiation is its retrieval-and-citation layer; ChatGPT’s is its breadth of generation and workflow tools. | Users who need a general assistant across writing, coding, analysis, and creative tasks, and who expect a $20 subscription to cover professional as well as personal work. | Free tier available. Plus $20/mo (GPT-6, higher message and upload limits, image generation, enhanced Deep Research, Codex, Projects). Pro from $100/mo. Business and Enterprise tiers available. |
| Google Gemini (Google AI Pro) Google AI plans (2026 restructuring) Gemini top-tier model access restricted to paid plans | Google’s multimodal assistant, deeply integrated with Google Search, Workspace, and the broader Google ecosystem. | Google AI Pro is $19.99/month and includes Gemini’s top-tier model, Deep Think reasoning, and integration with Google services (Gmail, Docs, Drive, YouTube). It is positioned as a general assistant within Google’s ecosystem rather than a citation-first answer engine. Commercial-use terms differ from Perplexity’s consumer restriction; Google AI Pro is sold as a productivity subscription. Perplexity’s answer surface is explicitly citation-centric; Gemini’s is multimodal and ecosystem-integrated. | Users already inside the Google ecosystem who want an assistant integrated with Gmail, Docs, and Search, and who value multimodal reasoning over citation-first retrieval. | Free tier available. Google AI Plus $7.99/mo. Google AI Pro $19.99/mo (top-tier Gemini model, Deep Think, 5 TB storage). Google AI Ultra from $99.99/mo. API pricing separate. |
Comparison checked against official product information on 2026-11-05. Features and pricing may change.
Frequently Asked Questions
Can I use Perplexity Pro for commercial work?
No. Perplexity’s consumer Terms of Service restrict Free, Pro, and Max to “personal, non-commercial use only” and prohibit exploiting the Services for any commercial purpose. A $20/month Pro subscription does not authorize client work, revenue-generating research, or internal business use. Commercial use requires Enterprise Pro ($40/seat/month) or Enterprise Max ($325/seat/month), which are governed by separate Enterprise terms.
How many Pro Searches does Perplexity Pro include?
Perplexity does not publish a specific number for consumer Pro. The official plan-comparison table lists Pro Search on Pro as “weekly limits (average use).” Enterprise Pro is listed at 400 per week and Enterprise Max at 4,000 per week. Third-party analyses and user reports place the consumer Pro effective range between roughly 100 and 300 per week, but Perplexity has not confirmed an exact figure. The limit is described as an average rather than a guaranteed per-day allowance.
What is the difference between Perplexity Pro and the API?
Pro is a web and app subscription for personal use. The API Platform is a separate pay-as-you-go system for developers, billed per token and per request, with no shared credits and no access to Pro features in the web interface. API pricing is token-based: Sonar starts at $1 per million input and output tokens with a $5–$12 per 1K request fee; Sonar Pro is $3 input / $15 output with a $14 per 1K request fee. A Pro subscription does not include API credits, and API spend does not count toward Pro allowances.
Is Comet free?
The Comet browser itself is available at no cost. Comet Plus, a separate AI-powered news and content service, is $5/month for users without a paid Perplexity plan and is included with Pro and Max. Comet Assistant queries are subject to plan limits: Free has no included Assistant queries, Pro has “monthly limits (average use),” Max has “monthly limits (advanced use),” Enterprise Pro includes 80 per month, and Enterprise Max includes 800 per month.
Does Perplexity train on my data?
On Free, Pro, and Max, training is the default unless you opt out; the plan-comparison table lists “Opt-out AI training available” for Pro and Max. On Enterprise Pro and Enterprise Max, data is “never used for model training” as a contractual default. The API Platform also states that data is never logged or used for training. Users handling confidential material on a consumer plan must enable the opt-out rather than rely on a default exclusion.
Alternative Tools
-
NotebookLM NotebookLM (now Gemini Notebook) is Google's source-grounded research assistant: you upload documents, and it answers only from those sources with inline citations, then turns them into Audio Overviews, Video Overviews, Mind Maps, Slide Decks, and study tools. The free tier is genuinely usable; paid limits are bundled inside Google AI plans. The hard boundaries are no consumer API, answers bounded by uploaded sources, and Audio Overview language quality that is still catching up outside English. -
Elicit Elicit is a systematic review workflow tool with a strong structured extraction interface, but independent evaluations find its search sensitivity averages 39.5% and its quote/reasoning matching degrades to 10%/0% in high-accuracy mode, positioning it as a complement to human reviewers rather than a replacement. -
Exa Exa is a semantic search API built on its own 500B+ page index, with page text included in search results, strong entity-lookup accuracy, an official MCP server, and per-request pricing from $4/1k. It is not a benchmark leader across the board—it is statistically tied with Brave and Firecrawl on independent agentic search benchmarks—and its 10 QPS limit, non-paginating search, crawl-bounded recency, and Enterprise-only compliance features are hard constraints that require workflow design around them. -
Tavily Tavily is a web access API for AI agents with a genuinely usable free tier (1,000 credits/month, no card), bundled full-page text in search results, and an official MCP server. But its Research endpoint is dynamically priced at 4–250 credits per request, the headline 1,000 RPM applies only to paid Production keys, entity and academic retrieval lag specialized indexes by large margins, and query data may be used for training unless Zero Data Retention or the data-use setting is changed. -
Brave Search API Brave Search API provides web, news, image, video, local, and LLM Context results from Brave's independent index, priced per request with a $5 monthly credit, aimed at AI agents and RAG pipelines rather than Google SERP replication. -
ChatGPT ChatGPT is a general-purpose AI assistant for writing, file and data analysis, web research, image generation, coding, and voice conversations.