ChatGPT Search vs Perplexity: Sources, Citations, Plans and Data Use Side by Side

ChatGPT search and Perplexity both answer questions with text written from live web results, but OpenAI and Perplexity describe their retrieval, citations, plans, ads and data settings in different terms, which makes a like-for-like comparison hard. This page puts the documented facts for both products side by side: how each one searches and cites, which models each names, what each plan includes, where ads appear, how chats are used for training, what business versions offer and what publishers can control. It is written for individuals choosing an answer engine, IT and privacy teams deciding what staff may use, and publishers who want to know how each product reaches their pages. Products are listed alphabetically (ChatGPT search, then Perplexity) and order is never a ranking. Every fact comes from OpenAI's or Perplexity's own pages, linked in the sources list.

Last checked: 2026-11-15.

This is an independent reference page from NHI Mgmt Group based on public provider documentation; it is not sponsored or endorsed by any provider. Both products change often, so confirm plans, limits and settings on each provider's current pages.

Key takeaways

• Both cite sources inside the answer. OpenAI describes inline citations plus a Sources button that lists cited and related links; Perplexity states that each answer includes numbered citations linking to the original sources.
• Retrieval works differently on paper. OpenAI says disassociated search queries may be shared with third-party search providers such as Bing; Perplexity runs its own crawler, PerplexityBot, to surface and link websites in its results.
• Free tiers exist with limits. OpenAI lists Search on ChatGPT Free with limited deep research; Perplexity lists 3 Pro Searches per day and 1 Research query per month on its Free plan.
• Consumer training opt-out, business no-training. OpenAI marks "Opt-out available" on Free, Go, Plus and Pro and states it does not train on business data; Perplexity lists opt-out on Pro, Education Pro and Max and "Data never used for model training" on its Enterprise plans.
• Publishers control each product with a named crawler. Blocking OAI-SearchBot removes a site from ChatGPT search answers (navigational links aside); Perplexity recommends allowing PerplexityBot for a site to appear in its results.

What are ChatGPT search and Perplexity?

ChatGPT search is a capability inside ChatGPT, OpenAI's assistant app, that lets ChatGPT search the web and return answers with links to web sources. OpenAI's help centre describes it as offering "fast, timely answers with links to relevant web sources", so a user does not need a separate search engine. It runs on the ChatGPT website, desktop apps and mobile apps, and it can be triggered by the user or chosen automatically by ChatGPT.

Perplexity is an AI answer engine from Perplexity AI. Its help centre says it searches the internet, gathers information from sources such as articles, websites and journals, and writes a conversational summary with numbered citations. Search is the core product rather than one feature of a wider assistant, and Perplexity sells it in individual and enterprise plans, alongside a separate API platform and its Comet browser assistant.

The deeper single-product profiles are ChatGPT Search Explained and Perplexity AI Explained. This page covers only the points where a reader needs both in one view.

The two products at a glance

How does each product find and cite sources?

ChatGPT search

• Triggering: a user clicks the web search icon in the message box, types "/" and picks Search, or lets ChatGPT search automatically when a question could benefit from web information.
• Retrieval: searches are based on the user's prompts, and OpenAI states that disassociated search queries may be shared with third-party search providers such as Bing. OpenAI's own crawler, OAI-SearchBot, surfaces websites in ChatGPT's search features.
• Location: ChatGPT collects general location based on the user's IP address and may share it with search providers to improve result accuracy.
• Citations: responses that use search include inline citations; on desktop web, hovering shows more detail and clicking opens the source. A Sources button at the end lists all cited sources and other relevant links. Some results show images at the top, each with its own citation and source link.
• Referral tagging: ChatGPT adds utm_source=chatgpt.com to referral URLs, so publishers can see ChatGPT traffic in their analytics.

Perplexity

• Triggering: every query is a search; users can choose Pro Search for more thorough research across multiple sources, or Research for longer reports.
• Retrieval: Perplexity's help centre says it searches the internet and gathers information from authoritative sources. Its documentation names two agents: PerplexityBot, "designed to surface and link websites in search results on Perplexity", and Perplexity-User, which fetches a page when a user asks a question.
• Location: Not stated by Perplexity on the pages checked.
• Citations: each answer includes numbered citations linking to the original sources, so readers can verify the information or explore further.
• Referral tagging: Not stated by Perplexity on the pages checked.

The practical difference for readers is the same in both: the answer is a summary, and the cited pages are the evidence. Opening the cited source before acting on a figure or claim is the habit both help centres support by putting links next to the text.

Which models does each product use?

As of 2026-11-15. Perplexity's help article on how it works names GPT-5 and Claude 4.6 Sonnet, while its plan page names GPT-6 Astra and Claude Opus 5.5 for Enterprise Max, so the model list differs by page and plan. Model detail is on the Frontier AI Models Compared page.

How much does each cost, and what are the limits?

Neither provider's pricing page, as fetched on 2026-11-15, showed every price in plain text. The table lists what each provider states; where a price is not shown, the cell says so for both.

Availability: OpenAI notes that Enterprise and Business can purchase credits for more access. Perplexity notes that Max annual billing is available only on perplexity.ai and that eligible educational, nonprofit and government organisations can get a 25% discount on Enterprise seats. Check each provider's live checkout for the price in your currency.

Do ads appear in the answers?

ChatGPT search

OpenAI announced on 16 January 2026 that it would test ads on the Free and Go tiers, starting with logged-in adults in the US, and that Plus, Pro, Business and Enterprise subscriptions would not include ads. Its stated principles:

• Answer independence: "Ads do not influence the answers ChatGPT gives you."
• Placement: ads appear at the bottom of answers, only when a relevant sponsored product or service matches the conversation, and not near sensitive or regulated topics such as health, mental health or politics.
• Labelling: ads are clearly labelled and separated from the organic answer.
• Privacy: conversations are kept private from advertisers, and OpenAI says it never sells data to advertisers.
• Age: no ads in accounts the user indicates, or OpenAI predicts, belong to someone under 18.
• Controls: users can see why they are shown an ad, dismiss it, turn off personalisation and clear data used for ads.

OpenAI's pricing page marks the Go plan "This plan may include ads". Market-by-market expansion is tracked on the Ads in AI Answers Tracker.

Perplexity

Perplexity's July 2024 Publishers Program post described its advertising model: brands can pay to place related follow-up questions in the answer engine interface and on Pages. Its plan comparison page, as checked on 2026-11-15, does not list ads as a feature or limit of any plan, so readers who need the current position should check Perplexity's own announcements.

Are searches used for training, and what can users control?

For work use, consumer accounts on either product leave the training choice with each individual. Organisations that need a guaranteed position should use the business tiers and set policy for personal accounts; see the Blocking Personal ChatGPT, Claude and Gemini Accounts at Work guide and the Safe Use of AI Chatbots Guide.

Business and enterprise versions

ChatGPT search

• Plans: ChatGPT Business, ChatGPT Enterprise and ChatGPT Edu.
• Training: "We do not train on your business data, and our models don't learn from your usage."
• Retention: customer-controlled retention on ChatGPT Enterprise.
• Security: AES-256 encryption at rest, TLS 1.2 or higher in transit, SOC 2 audit, enterprise SAML SSO.
• Admin: fine-grained control over access and available features.

Deployment detail is in the ChatGPT Enterprise Security Guide.

Perplexity

• Plans: Enterprise Pro and Enterprise Max.
• Training: "Data never used for model training"; data is never logged or used for training.
• Retention: configurable Data Retention on Enterprise Max, where the organisation has at least one Enterprise Max user.
• Security: SCIM and Audit Logs on Enterprise Max, under the same condition.
• Admin: billing and account access management, advanced seat management, Shared Projects, centrally governed File Connectors and an Insights dashboard. The Organization File Repository is being retired on 1 October 2026.

Deployment detail is in the Perplexity Enterprise Security guide.

What can publishers control?

Publishers who want to appear in one product and not train the other's models can set these rules independently. The wider set of crawler rules is covered in AI Crawler and Scraper Controls.

How to set publisher rules step by step

1. Decide per goal: appear in answers, allow user-triggered fetches, allow training. These are separate decisions for each provider.
2. Write robots.txt groups for OAI-SearchBot and PerplexityBot (search) and GPTBot (training) to match the decision.
3. Test on a staging path first: confirm the file parses and that only the intended paths are disallowed.
4. Verify crawler identity before allowing or blocking by user agent at the edge; Perplexity publishes IP lists in perplexitybot.json and perplexity-user.json.
5. Monitor: watch server logs for each user agent and analytics for chatgpt.com referrals after about 24 hours.
6. Roll back by restoring the previous robots.txt; changes are picked up on the next crawl.

How to choose: factual criteria

The right product depends on the use, not on a ranking. These criteria come from the documented facts above.

• You already use ChatGPT for other work: ChatGPT search sits in the same conversation, with the same account, memory settings and plan.
• You want to choose the model for each search: Perplexity documents a model selector in Pro Search with models from several developers; ChatGPT documents a model picker by plan with OpenAI models.
• You need an ad-free experience: OpenAI states Plus, Pro, Business and Enterprise do not include ads; Perplexity's plan pages do not list ads, and its 2024 Publishers Program post described sponsored follow-up questions.
• You need published usage numbers for budgeting: Perplexity publishes weekly and monthly counts for Enterprise Pro and Enterprise Max; OpenAI describes deep research as limited, expanded or maximum by plan.
• You need SCIM, audit logs and retention control: OpenAI lists SSO and retention control on ChatGPT Enterprise; Perplexity lists SCIM, Audit Logs and configurable retention on Enterprise Max.
• Small team versus large organisation: a small team can start on business entry tiers (ChatGPT Business, Perplexity Enterprise Pro) and move up when it needs identity lifecycle and retention controls.

Measuring use and impact

• Publisher referrals: count sessions tagged utm_source=chatgpt.com and referrals from perplexity.ai in analytics, month on month.
• Crawler activity: requests per day from OAI-SearchBot, PerplexityBot, ChatGPT-User and Perplexity-User in server logs.
• Workforce use: share of AI search traffic going to managed business accounts versus personal accounts, from proxy or CASB logs.
• Policy coverage: percentage of staff on business plans with training excluded by default.
• Citation checks: sample of answers used in published work where the cited source was opened and confirmed.

Common failure patterns

• Blocking the training crawler and expecting to leave search: GPTBot and OAI-SearchBot are independent, so blocking only GPTBot keeps a site eligible for ChatGPT search.
• Expecting robots.txt to stop user-triggered fetches: both providers say their user-triggered agents may not follow, or generally ignore, robots.txt.
• Treating consumer and business plans as the same: training defaults differ by tier on both products.
• Quoting the summary instead of the source: both products show citations so the reader can check the original page.
• Relying on an old plan table: both providers have changed plan names, limits and models during 2026.

Where identity and security fit

Both products are reached through user accounts, and business tiers add single sign-on and, on Perplexity Enterprise Max, SCIM. Tie business accounts to the corporate identity provider, remove access on leaving, and treat any API keys for either provider's API as secrets. See the AI model provider API keys guide for key handling, and the Agentic Browsers Security Guide for browser agents such as Comet.

Practitioner checklist

• Decide which product, if any, staff may use for work, and on which tier.
• Move work use to business tiers where training is excluded by default, and connect them to SSO.
• For personal accounts that remain, publish how to use the training opt-out on each product.
• Write separate robots.txt rules for search crawlers and training crawlers, then verify with server logs.
• Track chatgpt.com and perplexity.ai referrals in analytics as a baseline.
• Check ad exposure: ChatGPT Free and Go may show ads; record Perplexity's current position from its own pages.
• Recheck plans, models and limits on both providers' pages every quarter.

Standards and references

• OpenAI Help Center: ChatGPT search
• Perplexity Help Center: How does Perplexity work?
• OpenAI: ChatGPT pricing
• Perplexity Help Center: Which Perplexity subscription plan is right for you?
• OpenAI: Overview of OpenAI crawlers
• Perplexity: Perplexity crawlers
• OpenAI: Our approach to advertising and expanding access
• OpenAI: Enterprise privacy
• OpenAI Help Center: Publishers and developers FAQ
• Perplexity Help Center: What data does Perplexity collect about me?
• Perplexity: Introducing the Perplexity Publishers Program

Related NHI Mgmt Group resources: AI Answer Engines Hub · ChatGPT Search Explained · Perplexity AI Explained · Google AI Mode vs ChatGPT Search