AI Search Engines List: Every Major Engine Compared

This list compares engines from a website owner's side — how each one finds, cites and sends visitors to pages — not which assistant is best to use.

Which AI search engines cite web sources?

Seven AI search engines cite web sources with links: Google AI Overviews, Google AI Mode, ChatGPT search, Perplexity, Gemini, Microsoft Copilot and Claude. Each one retrieves pages from a documented index or search provider and shows the pages it used beside or inside the answer.

An AI search engine is a system that answers a question in its own words and links the web pages it drew on. For a website owner, four attributes decide how a site reaches each one: the index source the engine retrieves from, the crawler token that controls access in robots.txt, the citation style that shows your page to the reader, and the referral visibility that decides whether a click can be identified in analytics. A fifth column records whether the operator gives site owners a report of their own.

Source keys: [AIF] Google, AI features and your website · [GSC] Search Console Help, Generative AI performance report · [GCC] Google, Common crawlers · [GAH] Gemini Apps Help, related sources · [OAB] OpenAI, crawlers overview · [OAH] OpenAI Help, Searching the web with ChatGPT · [OAF] OpenAI Help, Publishers and Developers FAQ · [PB] Perplexity, crawlers · [PH] Perplexity Help, How does Perplexity work · [MSC] Microsoft Support, Copilot in Bing and Responsible AI · [BHD] Microsoft Support, How Bing delivers search results · [BAP] Bing Webmaster Blog, AI Performance · [ACB] Anthropic, crawler article · [AWS] Claude API, web search tool. All checked 25 September 2026; full links in the Sources list below.

The table’s main finding is that the engines agree on the least important attribute for site owners — all of them link sources — and differ on the ones that decide access and measurement. The gaps are as informative as the values: 7 of the 49 cells read “Not documented”, all of them in the referral and reporting columns.

Grouped by operator, the 7 rows fall into five families:

• Google (3 engines): AI Overviews, AI Mode and Gemini — one index, one crawler, one control token that affects only Gemini.
• OpenAI (1): ChatGPT search — its own search crawler and the only documented referral tag.
• Perplexity (1): its own crawler plus a user-triggered fetcher.
• Microsoft (1): Copilot — the Bing index and the second first-party report.
• Anthropic (1): Claude’s API web search — three crawler tokens, provider not named.

Where does each AI search engine get its pages?

AI Overviews, AI Mode and Gemini draw on the Google Search index; Copilot grounds in Bing web results; ChatGPT uses third-party search providers and partner content; Perplexity runs real-time web search with its own crawler; Claude's API runs web searches through an unnamed provider.

The index source is the collection of pages an engine retrieves from before it writes an answer. It decides the first question for any site: which index do I need to be in? The seven engines fall into three groups.

• The Google Search index: AI Overviews and AI Mode link pages that are indexed and eligible for a snippet. Gemini uses the same index through grounding, which Google defines as “providing content from the Google Search index to the model at prompt time to improve factuality and relevancy.”
• The Bing index: Microsoft says Copilot in Bing is “grounded in web search results”: the system runs one or more web searches and uses the top results to write its summary.
• Engine-run or third-party web search: OpenAI says ChatGPT search “leverages third-party search providers, as well as content provided directly by our partners”, without naming the providers. Perplexity searches the web in real time and runs its own crawler, PerplexityBot. For Claude’s API, Anthropic states that “the API runs the searches and provides Claude with the results”, and does not name the search provider.

For a site owner, the consequence is that being indexed in Google covers three of the seven engines at once, and being indexed in Bing covers a fourth. The other three depend on their own crawlers or on providers their operators do not name.

Which crawler token belongs to which AI engine?

Googlebot serves AI Overviews, AI Mode and Gemini grounding; OAI-SearchBot serves ChatGPT search; PerplexityBot serves Perplexity; Bingbot serves Copilot; Claude-SearchBot serves Claude search. Training crawlers and user-triggered fetchers are separate tokens with separate robots.txt rules.

A crawler token is the name a crawler answers to in robots.txt, the file that tells crawlers which paths they may fetch. The tokens that matter for AI answers fall into four roles:

• Search crawlers, which decide whether a page can be cited: Googlebot, OAI-SearchBot, PerplexityBot, Bingbot and Claude-SearchBot.
• User-triggered fetchers, which open a page when a user’s question calls for it: ChatGPT-User, Perplexity-User and Claude-User. OpenAI says robots.txt rules “may not apply” to ChatGPT-User, and Perplexity says Perplexity-User “generally ignores robots.txt rules”; Anthropic states that disallowing Claude-User stops Claude retrieving the site for user queries.
• Training crawlers, which collect content for model training and do not decide search citations: GPTBot and ClaudeBot.
• Control tokens, which change how content crawled by another bot is used: Google-Extended. It has no user agent of its own and does not affect Google Search, AI Overviews or AI Mode; it governs Gemini training and Gemini Apps grounding.

The rule that follows is simple: blocking a search crawler removes a site from that engine’s citations, while blocking a training crawler does not. A single User-agent: * disallow removes all of them at once.

How does each engine display its citations?

Every engine in this list links its sources, in 3 forms: links inside Google's AI features, inline citations with a sources panel in ChatGPT and Gemini, and clickable citations or references in Perplexity and Copilot. Claude's API returns a cited text excerpt of up to 150 characters.

Citation style is how a source appears to the reader, and it shapes how often a citation turns into a visit. Google describes pages in AI Overviews and AI Mode as “supporting links”. OpenAI says ChatGPT search cites “via inline citations” and a Sources button that opens a sidebar of references. Gemini Apps show “the Sources button at the bottom of the response or in-line throughout the response”, which opens a side panel, though Google notes that “not all responses include related links or sources”. Perplexity’s help center says every answer comes with clickable citations to the original sources, and Microsoft says Copilot’s summaries include references to the search results used.

Claude’s API works differently because developers build the interface. Anthropic states that “citations are always enabled for web search”, and each citation returns the URL, the page title and “up to 150 characters of the cited content”. How those citations look to a reader depends on the application that displays them.

Which AI engines send traffic you can identify in analytics?

ChatGPT is the only engine in this list whose operator documents a referral tag: it adds utm_source=chatgpt.com to its links. Google counts AI feature clicks inside Search Console's Web search type; the other operators document no tagging, so their visits depend on the browser referrer.

Referral visibility is whether a visit from an engine can be told apart from other traffic. A UTM parameter is a tag added to a link’s URL that analytics tools read as the visit’s source; a referrer is the address of the previous page, which the browser passes along when the visitor clicks.

OpenAI’s Publishers and Developers FAQ says ChatGPT includes utm_source=chatgpt.com in referral URLs, so ChatGPT visits carry their own label in GA4. Google takes a different route: AI Overviews and AI Mode clicks are counted in Search Console’s main Performance report under the “Web” search type, mixed with regular results. For Gemini, Perplexity, Copilot and Claude, the operators document no tag of their own. Visits from those engines can only be identified from the referrer where the browser passes one, and any visitor who copies a URL instead of clicking arrives as direct traffic.

Which AI engines report citations to site owners?

Two operators report AI citations to site owners: Google, through Search Console's Generative AI performance report for AI Overviews and AI Mode, and Microsoft, through Bing Webmaster Tools' AI Performance report for Copilot. OpenAI, Perplexity and Anthropic document no equivalent report.

Google’s report counts impressions — how many times links to a site were shown in a generative AI feature on Google Search — by page, country, device and date, and has been available to all sites since 31 August 2026. It covers AI Overviews and AI Mode, not the Gemini app. Microsoft’s AI Performance report, in public preview since 10 February 2026, covers “Microsoft Copilot, AI-generated summaries in Bing, and select partner integrations”, with total citations, average cited pages and grounding queries.

For the other four engines, measurement means running a fixed set of prompts on a schedule and logging the cited pages yourself.

The table says what each engine is. Getting cited by one of them is a separate job, covered engine by engine.

Optimization guides by engine

Each guide covers one engine in depth: its crawler, its source selection and the steps that raise citations there.

• AI Overviews citation guide

  Google's AI summaries at the top of search results. Three eligibility checks, four snippet controls and the Search Console report.

• AI Mode source selection

  Google Search's conversational mode. How query fan-out picks pages, and how to cover the sub-queries.

• Gemini SEO

  Google's standalone AI assistant. Grounding in the Search index and what the Google-Extended token controls.

• ChatGPT SEO

  OpenAI's search feature in ChatGPT. OAI-SearchBot access, 6 steps to a citation and the GA4 path for tagged referrals.

• Perplexity SEO

  An answer engine with its own crawler. PerplexityBot access, answer-ready passages and referral tracking.

• Guide planned

  Microsoft Copilot

  Microsoft's AI assistant, grounded in Bing web results. Guide planned.

• Guide planned

  Claude

  Anthropic's AI assistant, with three crawler tokens and API web search citations. Guide planned.

Guides for Meta AI, Grok, DeepSeek and Apple Intelligence are planned as well. Each engine joins the comparison table only after its operator’s own documentation has been checked for all five attributes.