ChatGPT, Gemini, Perplexity, Claude: Why every AI system talks about you differently

Key takeaways

• ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews answer the same question in different ways. Each one searches in its own way, draws on its own sources and displays those sources with varying prominence.
• In our GeoUp measurement series, Perplexity mentions the companies studied in 17.0% of answers and is the strongest system for six of nine visible client websites. Claude, at 4.4%, offers the greatest potential.
• Providers update their models at short intervals. Between September 22 and October 7, 2026 alone, Anthropic, Google and OpenAI introduced new models.
• Measuring your visibility per system shows you where you are already strong and lets you focus your work where there is the most to gain.

As of October 2026

A managing director wants to know how her company is doing in AI answers. She asks ChatGPT for recommended providers in her industry and finds her company in the answer. Her head of sales asks Perplexity the same question and gets a slightly different list, with different sources beneath it. Both answers are plausible. They simply come from different systems.

This comes down to how the systems are built, and it is good news. Visibility in ChatGPT, Gemini and Perplexity comes about in different ways, and you can use each of these ways deliberately.

Four reasons why AI systems answer differently

Behind every AI answer are decisions that each system makes differently: whether it searches the web, where it searches and how clearly it shows its sources. Then there is the language model that writes the answer and is regularly replaced by a newer one. These four differences explain why a company is mentioned particularly often in one system and still has room to grow in another.

Difference 1: When the systems search the web live

Perplexity is built as an answer engine. In our measurements, it starts every answer with a live web search and builds the answer from the sources it finds. For this, Perplexity runs its own crawler, a program that retrieves websites automatically: PerplexityBot is designed to surface and link websites in Perplexity's search results.

Gemini and ChatGPT decide, depending on the question, whether a search is worthwhile. In its developer documentation, Google describes how Gemini determines on its own whether a Google Search would improve the answer and then runs one or more search queries. OpenAI's models can also search multiple times and decide whether to keep searching.

Google works with many sub-searches in its own search engine: AI Mode breaks a question down into subtopics and researches all of them at the same time. To learn how this relates to classic search engine optimization, read our article on the SEO foundation of AI visibility.

Claude can search the web as well. According to Anthropic, Claude searches when doing so helps the answer. In our measurement series, Claude answered without a live web search, drawing on the knowledge the model acquired during training.

What this means for you: systems with live search pick up improvements to your website quickly. A model's training knowledge grows over a longer period, with every mention of your company on the web.

Difference 2: Where the systems get their sources

Every provider searches the web with its own technology. Google relies on its search index. OpenAI runs the OAI-SearchBot crawler, which makes websites visible in ChatGPT's search features. Anthropic collects content for search in Claude with Claude-SearchBot, and Perplexity with PerplexityBot.

This leads to a simple and effective step: give the search crawlers of all systems access to your website. OpenAI and Perplexity explicitly recommend allowing their crawlers in robots.txt, and Perplexity also advises allowing them in any upstream firewall. A quick look at your robots.txt and firewall settings shows whether all paths are open.

Beyond your website, the systems draw on many other sources. In our analysis of 12,769 AI answers, 56% of the answers linked to sources. At the top of the list were YouTube, business directories, Reddit, trade media, press portals and Wikipedia.

Difference 3: How prominently the sources appear in the answer

The systems also take their own approach to displaying sources:

• Perplexity links the sources it finds visibly in the answer.
• Claude backs up answers that use web search with citations and links to the sources.
• Gemini shows sources via a button at the end of or within the answer, although not every answer contains links.
• OpenAI provides inline citations with the source's URL and title by default when its models search the web.
• Google adds links to further information to its answers in AI Mode.

Both forms of visibility have value. Where the AI links to your website, prospects reach you with a single click. Where it mentions your company without a link, your name sits right in the middle of the recommendation.

Difference 4: New models in fall 2026

The models behind the systems are updated at short intervals. In the past few weeks alone:

• September 22, 2026: Anthropic releases Claude Opus 5.5, the first model in the new Claude 5.5 family.
• September 30, 2026: Google introduces Gemini 4 Argon. It goes first to selected IT security teams, then to paying API customers and Google AI Ultra subscribers.
• October 7, 2026: OpenAI brings GPT-6 Sol and GPT-6 Luna to all ChatGPT users, on free and paid plans.

A new model may phrase things differently, weight things differently and search in different situations. In our measurements, individual companies' scores shift noticeably within a single week. At the same time, visibility can be built up deliberately: a regional service provider increased its presence in AI answers from 5.0% to 14.2% within two months and held that level in the following measurements.

What our GeoUp measurement series shows for each system

With our tool GeoUp, we continuously measure how often AI systems mention companies. For our September 2026 analysis, we compared 16 client websites, using the average of the last three measurement runs per website and around 1,200 answers per system (as of September 25, 2026). Each measurement run asks five AI systems 30 questions. This is how often the systems mentioned the companies studied:

• Perplexity: 17.0% of answers
• ChatGPT: 11.3%
• Gemini: 9.6%
• Google AI Overviews: 7.2%
• Claude: 4.4%

This makes Perplexity the strongest system for many of our clients. For six of nine visible client websites, it accounts for the most mentions. A kitchen appliance manufacturer appears there in 47.8% of answers, an automotive service provider in 46.7%. For two websites, ChatGPT is the strongest system, once tied with Perplexity, and for one well-known brand it is Claude.

Claude offers the greatest potential. Because Claude answers from its training knowledge in our measurement series, visibility there grows through mentions outside your own website: in the press, directories, trade media and Wikipedia. These are exactly the sources that the systems with live search frequently cite as well, so working on them pays off several times over.

Even in our July analysis, the most visible system and the most reserved one were a factor of 4.5 apart. That is why only a breakdown by system shows where your visibility comes from. GeoUp reports the results for each system separately. With Market Battle Mode, you can also see your visibility in direct comparison with your competitors across all five systems, including the sources the AI cites for them.

How to set the right priorities for each system

1. Measure per system. The starting point is a measurement that reports each system separately. GeoUp asks ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews 30 typical customer questions from your industry and records whether your company is mentioned, in which position and which sources the AI cites. The measurement runs regularly, up to weekly, and you can add your own questions.
2. Build on your strengths. In systems with live search, content on your website takes effect quickly. Citable pages on the questions your customers actually ask have the greatest impact there.
3. Tap into your potential. For the models' training knowledge, mentions in the press, trade media and directories are what count. Genuine occasions such as a new project or an award are particularly well suited for this.
4. Open all paths. Allow the search crawlers from Google, OpenAI, Anthropic and Perplexity, and add complete, correct schema markup so that every system can clearly attribute your information to your company.
5. Measure again after model changes. When a provider introduces a new model, the next measurement shows how your scores are developing in that system.

This work is easy to divide up: GeoUp measures, our team implements. As part of our AI visibility service, we regularly write new articles on exactly the questions your customers ask AI, strengthen your presence on your website and wherever AI systems look things up, and show you every month what has become more visible. You can book our SEO and GEO expert team at any time in addition to GeoUp. Simply write to us at our email address.

Frequently asked questions

Which AI system matters most for my company?
That depends on your industry and how well known you are, which is why measuring per system is worthwhile. In our measurement series, Perplexity is the strongest system for six of nine visible client websites. For one well-known brand, Claude came out ahead.

Why does Perplexity mention me more often than Claude?
In our measurements, Perplexity searches the web live for every answer and so accesses your website directly. In our measurement series, Claude answered from its training knowledge, which grows through mentions in the press, trade media and directories. That is where the greatest potential lies for many companies.

Do new models such as GPT-6 or Gemini 4 change my visibility?
A new model may search differently and weight sources differently, so it is worth checking your scores after a model change. By maintaining your website and your sources, you create a foundation that every new model can draw on as well.

Is it enough to monitor only ChatGPT?
ChatGPT is an important channel and shows part of the picture. In our measurement series, Perplexity mentioned the companies studied in 17.0% of answers, ChatGPT in 11.3%. Measuring all five systems shows you which system carries your visibility the most.

Conclusion

ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews each search, weight and cite in their own way, and their models change at short intervals. For companies, this is an opportunity, because each system opens up its own path into the answer. Systems with live search pick up a good website quickly, and training knowledge grows with every mention on the web. Measuring per system shows you where you are already strong and where the next step pays off most.

How visible is your company in each individual AI system?

With the free GeoUp quick check, you can see in moments, based on sample questions, whether ChatGPT, Perplexity, Claude, Gemini and Google AI Overviews mention your company. GeoUp is a product by njudev.

Would you rather hand off the implementation? Then our team will take care of it. You can book our SEO and GEO expert team at any time in addition to GeoUp. Simply write to us at our email address or learn more about our AI visibility service.

Sources

1. GeoUp: “Perplexity trägt die KI-Sichtbarkeit: Warum ein System den Unterschied macht”, September 17, 2026, measurement series as of September 25, 2026. https://geoup.io/news/perplexity-traegt-die-ki-sichtbarkeit (in German)
2. GeoUp: “12.769 KI-Antworten ausgewertet”, July 24, 2026. https://geoup.io/news/ki-sichtbarkeits-auswertung-12769-antworten (in German)
3. Perplexity: “Perplexity Crawlers”, Perplexity documentation. https://docs.perplexity.ai/guides/bots
4. Google AI for Developers: “Grounding with Google Search”, Gemini API documentation. https://ai.google.dev/gemini-api/docs/google-search
5. OpenAI: “Web search”, OpenAI API documentation. https://developers.openai.com/api/docs/guides/tools-web-search
6. Google: “Get AI-powered responses with AI Mode in Google Search”, Google Search Help. https://support.google.com/websearch/answer/16011537?hl=en
7. Anthropic: “Enable and use web search”, Claude Help Center. https://support.claude.com/en/articles/10684626-enable-and-use-web-search
8. OpenAI: “Overview of OpenAI Crawlers”, OpenAI API documentation. https://developers.openai.com/api/docs/bots
9. Anthropic: “Does Anthropic crawl data from the web, and how can site owners block the crawler?”, Claude Help Center, as of April 7, 2026. https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler
10. Google: “View related sources from Gemini Apps”, Gemini Apps Help. https://support.google.com/gemini/answer/14143489?hl=en
11. Anthropic: “Introducing Claude Opus 5.5”, September 22, 2026. https://www.anthropic.com/claude-opus-5-5
12. Google: “Gemini 4 Argon: our next era of frontier intelligence”, Google Blog, September 30, 2026. https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/
13. OpenAI: “GPT-6 Sol and GPT-6 Luna: October 2026 update”, OpenAI Deployment Safety Hub, October 7, 2026. https://deploymentsafety.openai.com/gpt-6-october
14. GeoUp: “Market Battle Mode”. https://geoup.io/market-battle (in German)