Grokipedia vs Wikipedia: Quantitative Analysis (video)

Recently, Grokipedia v0.1 was released — an AI-based knowledge system inspired by Wikipedia. But how complete is it? And what does its topic coverage look like compared to Wikipedia? Moreover, how does it differ across the many language editions of the world’s largest encyclopedia?

Wikipedia exists in hundreds of languages, but the topics covered vary widely from one edition to another. Some language communities focus more on science, others emphasize culture, geography, or biographies. These editorial and cultural priorities shape the way knowledge appears in each language.

In this breaf analysis, we look at 16 language editions of Wikipedia and Grokipedia, comparing:

• the number of articles across various topics
• the share of each topic within every knowledge base

All Wikipedia data come from the official dumps dated 1 November 2025, ensuring up-to-date and consistent results.

General Statistics

Let’s start from general statistics – total number of articles in 16 selected language versions of Wikipedia and Grokipedia v0.1:

• English Wikipedia (en.wikipedia.org): 7,084,648
• German Wikipedia (de.wikipedia.org) 3,065,206
• French Wikipedia (fr.wikipedia.org): 2,718,020
• Swedish Wikipedia (sv.wikipedia.org): 2,619,206
• Dutch Wikipedia (nl.wikipedia.org): 2,200,983
• Russian Wikipedia (ru.wikipedia.org): 2,069,606
• Spanish Wikipedia (es.wikipedia.org): 2,003,504
• Italian Wikipedia (it.wikipedia.org): 1,942,407
• Polish Wikipedia (pl.wikipedia.org): 1,673,513
• Chinese Wikipedia (zh.wikipedia.org): 1,508,209
• Japanese Wikipedia (ja.wikipedia.org): 1,478,676
• Ukrainian Wikipedia (uk.wikipedia.org): 1,395,976
• Vietnamese Wikipedia (vi.wikipedia.org): 1,296,664
• Arabic Wikipedia (ar.wikipedia.org): 1,284,243
• Portuguese Wikipedia (pt.wikipedia.org): 1,159,003
• Persian Wikipedia (fa.wikipedia.org): 1,060,282
• Grokipedia v0.1 (grokipedia.com): 885,279

Almost all articles in Grokipedia have corresponding articles in English Wikipedia. 24,288 article titles from Grokipedia were matched to corresponding Wikipedia articles through the redirect analysis. However, 3,536 Grokipedia articles were excluded from the analysis because there is no direct match to any title in English Wikipedia.

How Articles Were Assigned to Topics

Before diving into the results, let’s briefly explain how the articles were categorized. Often Wikipedia articles connected with corresponding Wikidata items, which contain such standardized properties as: “instance of” (P31), “country” (P17), “occupation” (P106). This allows us to identify what each article represents — whether it’s a city, a person, a company, a scientific concept, or something else.

However, Wikidata properties can be too specific or inconsistent for large-scale analysis. For example, one city might be described as a metropolis, another as a city-state, capital, or big city — yet they all belong to the same conceptual category. To standardize this, additionally a broader classification system was applied — the same is used on the WikiRank which organizes millions of Wikipedia articles into high-level topics such as: automobiles, books, films, buildings, cryptocurrencies, cities, programming languages, video games, biology, astronomy, and others (more information on this classification can be found in research studies [1], [2], [3]).

For example, let’s explore articles about Poznań. In Grokipedia, there’s a page titled “Poznań” describing the Polish city. If we check English Wikipedia, we can find article with the same title, and from there, we can easily switch to other language versions (e.q., Polish, German, French, Spanish) and find (different) titles related to that city. We can also easily find a corresponding Wikidata item Q268, which contains such statements as:

• “instance of”: “urban municipality of Poland” (WikiRank topic: cities)
• “instance of”: “city with powiat rights in Poland” (WikiRank topic: cities)
• “instance of”: “big city” (WikiRank topic: cities)
• “contry”: “Poland”
• and others…

Based on certain statements, we can align Grokipedia and Wikipedia article about about Poznań to such topics as “big city,” “urban municipality of Poland,” and “Poland,” as well as to the more general topic “cities”.

Due to space limitations and the large number of unique statements in Wikidata items, the analysis presents only the top 33 most frequently used values for each of the selected properties (related to all considered Grokipedia and Wikipedia articles from selected languages): “instance of” (P31), “country” (P17), “occupation” (P106).

Number of articles

The best results for each topic are highlighted in green.

Numbers: Wikidata property “instance of” (P31)

Numbers: Wikidata property “country” (P17)

Numbers: Wikidata property “occupation” (P106)

Numbers: WikiRank classification

Share of Topics

The values are given in percentages and represent the share of articles on a specific topic within the entire language edition (i.e., the number of articles on a given topic divided by the total number of articles). The highest results for each topic are highlighted in green.

Shares: Wikidata property “instance of” (P31)

Shares: Wikidata property “country” (P17)

Shares: Wikidata property “occupation” (P106)

Shares: WikiRank classification

Normalized results (video)

The video below shows normalized values using dynamic charts. Enjoy! 🙂

More information can be found on the website: wikipedia.lewoniewski.info

See also the dataset on Hugging Face and Kaggle that shows mapping Grokipedia pages to corresponding Wikipedia article titles across 16 languages.