LLM Citation Ranking by Bernstein Analytics (Q1 2026)

“Tagesschau” remains the most frequently cited source by language models, but the rankings behind it have shifted: welt.de has jumped from 11th to 2nd place.…

Published on: 12.06.2026
Reading time: 7 mins
Author: Carlo Hofacker

“Tagesschau” remains the most frequently cited source by language models, but the rankings behind it have shifted: welt.de has jumped from 11th to 2nd place. ChatGPT, Gemini, and Google’s AI Mode cite significantly more German media outlets than in the previous quarter, while Perplexity cites significantly fewer.

Third-Quarter Analysis: How We Conducted the Measurement

Which media outlets shape what AI systems say about Germany? For Bernstein Analytics’ third LLM citation ranking, we submitted 720 prompts on current topics in domestic politics, foreign policy, the economy, culture, and sports to five systems in the first quarter of 2026: ChatGPT, Gemini, Perplexity, Grok, and Google’s AI Mode. [Add number of topics.]

The 3,600 responses yielded 31,055 source references across 3,577 domains. 256 of these domains are traditional German media outlets. They account for 9,543 citations, or 30.7% of all sources.

Google AI Mode as a model. We included Google’s AI Mode because it is now displayed to a large number of users during Google searches, whether they want it to be or not. As a result, it reaches a very large number of people.

Here’s how the ranking is determined. The order follows our rating system: Each of the five models carries equal weight, regardless of how many sources it cites per response. A media outlet therefore performs well if multiple models cite it frequently, not just one.

Tagesschau Defends 1st Place, welt.de Moves Up to 2nd

tagesschau.de remains at the top. With 1,282 mentions, the portal accounts for 13.4% of all media citations; in the previous quarter, the figure was 10.9%. The lead is growing: The rating score rises from 12.3 to 18.1. Two systems are driving the gap. Google’s AI Mode mentions tagesschau.de 788 times, while Grok mentions it 264 times. With Gemini, however, the portal doesn’t appear even once.

The real movement lies behind these figures. welt.de jumps from 11th place to 2nd, placing it ahead of zeit.de, zdfheute.de, Deutschlandfunk, and Spiegel. With 668 mentions, all five models cite the portal. ChatGPT alone accounts for 348 of them. This puts it in first place for ChatGPT and makes it the sixth most-cited media source overall. zeit.de holds third place. Der Spiegel, which was still in second place in Q4, slips to sixth place.

To access the actual content, click the button below. Please note that doing so will share data with third-party providers.

Unblock content

More Information

Our rating rewards breadth. zdfheute.de ranks fourth, ahead of Deutschlandfunk and Spiegel, even though both have more mentions (573 and 612 compared to 558). The reason: The citations from zdfheute.de are distributed more evenly across the models. Spiegel, on the other hand, relies primarily on Gemini and AI Mode; it has 21 mentions in ChatGPT.

Notable Preferences of the Language Models

Gemini without Tagesschau. Gemini does not mention tagesschau.de, n-tv.de, t-online.de, or sportschau.de even once. All four are in the top 25 of the overall ranking. Instead, in Gemini’s rankings, *Der Spiegel* leads with 251 mentions (11.8% of media citations), ahead of Deutschlandfunk (241) and welt.de (208).

Grok and AI Mode: one in four sources comes from *Tagesschau*. For AI Mode, 26.8% of media citations come from tagesschau.de; for Grok, 25.9%. Grok distributes the remainder primarily across zdfheute.de (8.5%) and dw.com (8.3%).

ChatGPT relies on Welt and Zeit. welt.de (16.7%) and zeit.de (14.0%) together account for just under one-third of ChatGPT’s media citations. tagesschau.de reaches 7.7%.

Perplexity has the broadest distribution. No single media outlet exceeds 8%. Spiegel (7.8%), Zeit (7.1%), and Deutschlandfunk (6.0%) lead a tightly packed group at the top.

To access the actual content, click the button below. Please note that doing so will share data with third-party providers.

Unblock content

More Information

Four in, four out. Entering the top 10 are welt.de (from 11th place), bild.de (from 13th), tagesspiegel.de (from 18th), and dw.com (from 21st). Falling out of the top 10 are focus.de (from 5th to 13th), n-tv.de (from 6 to 19), fr.de (from 7 to 24), and merkur.de (from 10 to 28). merkur.de and fr.de suffered the steepest drops: they lost 18 and 17 spots, respectively.

Public broadcasting sites are making broad gains. sportschau.de enters the rankings at No. 11, swr.de climbs from No. 39 to 18, deutschlandfunkkultur.de from 29 to 15, mdr.de from 32 to 22, and ndr.de from 34 to 23. Zdfheute.de also moves up from No. 8 to 4. There are now four public broadcasting sites in the top 10 (tagesschau.de, zdfheute.de, Deutschlandfunk, dw.com); in Q4, there were three.

Focus and n-tv are losing ground. focus.de’s share of media citations drops from 3.7% to 2.1%, and n-tv.de’s from 3.9% to 1.5%.

To access the actual content, click the button below. Please note that doing so will share data with third-party providers.

Unblock content

More Information

AI Mode overtakes Grok. Google’s AI Mode accounts for 37.7% of German media across all sources (Q4: 19.8%) and replaces Grok (36.7%, Q4: just under 40%) at the top. ChatGPT rises from 23.6% to 33.6%, and Gemini from 19.7% to 29.8%. Perplexity drops from around 31% to 19.2%. Across all models, 30.7% of sources come from traditional German media.

For Grok and AI Mode, public broadcasting accounts for the majority. [Graph: Public broadcasting’s share of media citations, Datawrapper qVHQh] Of the media citations, 64.2% for Grok come from public broadcasting outlets (Q4: 36.0%), and 57.1% for AI Mode (Q4: 36.2%). ChatGPT increased its share from 5.2% to 21.2%, and Gemini from 22.9% to 24.7%. Only on Perplexity did the share decline slightly, from 25.0% to 23.5%. Across all models, 38.0% of media citations come from public broadcasting.

LLM Citation Ranking: Political Preferences

The political classification of media outlets is based on evaluations from professional and academic projects. A rating is available for 85.2% of media citations. All percentages refer to all media citations in the respective model; the remainder are classified as neutral or have no rating.

To access the actual content, click the button below. Please note that doing so will share data with third-party providers.

Unblock content

More Information

In four out of five models, left-liberal sources outnumber conservative ones. Perplexity leads with 36.8% left-liberal citations (conservative: 9.6%), followed by ChatGPT (27.5% to 20.6%), Grok (27.2% to 5.5%), and AI Mode (22.1% to 8.1%). Gemini is the exception: There, conservative sources, at 23.2%, are slightly ahead of left-liberal sources, at 22.9%.

To access the actual content, click the button below. Please note that doing so will share data with third-party providers.

Unblock content

More Information

The range between the models is widening. Conservative sources now account for 5.5% (Grok) to 23.2% (Gemini) of media citations; in Q4, the figures ranged from 12% to 16%. For left-liberal sources, the range extends from 22.1% (AI Mode) to 36.8% (Perplexity); in Q4, it was 22% to 29%.

The extremes. Perplexity sits at the left-liberal end with 36.8%, while Gemini is at the conservative end with 23.2%. Grok cites conservative sources the least frequently (5.5%), and AI Mode cites left-liberal sources the least frequently (22.1%).Compared to Q4, the conservative share in particular has shifted. For Gemini, it has risen by about 9 percentage points, and for ChatGPT by about 8. For Grok, it has fallen by about 10 points, for AI Mode by about 6, and for Perplexity by about 3. Left-liberal sources have increased by about 11 points for Perplexity and by about 5 for Grok, while they have lost just under 6 points for Gemini. ChatGPT and AI Mode remain virtually unchanged in terms of left-liberal sources.

Outlook

The next LLM citation ranking will be published in October 2026. The results from the previous quarter can be found in the Q4/2025 report.

About Bernstein Analytics GmbH

Bernstein Analytics GmbH supports its clients by systematically processing information for strategic decision-making. The company focuses on researching, analyzing, and evaluating data to accurately assess risks and opportunities in complex regulatory and communications environments. A core area of expertise is the analysis of narratives in AI language models (LLMs) as well as strategic consulting to optimize visibility within these systems.