LLMpedia: A Transparent Framework to Materialize an LLM's Encyclopedic Knowledge at Scale

TL;DR

LLMpedia demonstrates that large language models can generate a vast, open, and verifiable encyclopedic knowledge base entirely from their parameters, revealing gaps in current factuality benchmarks.

Contribution

This work introduces LLMpedia, the first fully open parametric encyclopedia, showcasing a scalable framework for materializing and evaluating LLMs' encyclopedic knowledge without retrieval.

Findings

LLMpedia produces approximately 1 million articles from LLMs without retrieval.

Verifiable true rate on Wikipedia subjects is 74.7%, lower than benchmark scores.

LLMpedia achieves higher factuality with less textual similarity compared to Wikipedia.

Abstract

Benchmarks such as MMLU suggest flagship language models approach factuality saturation, with scores above 90\%. We show this picture is incomplete. \emph{LLMpedia} generates encyclopedic articles entirely from parametric memory, producing ${\sim}$1M articles across three model families without retrieval. For gpt-5-mini, the verifiable true rate on Wikipedia-covered subjects is only 74.7\% -- more than 15 percentage points below the benchmark-based picture, consistent with the availability bias of fixed-question evaluation. Beyond Wikipedia, frontier subjects verifiable only through curated web evidence fall further to 63.2\% true rate. Wikipedia covers just 61\% of surfaced subjects, and three model families overlap by only 7.3\% in subject choice. In a capture-trap benchmark inspired by prior analysis of Grokipedia, LLMpedia achieves substantially higher factuality at roughly half the textual similarity to Wikipedia. Unlike Grokipedia, every prompt, artifact, and evaluation verdict is publicly released, making LLMpedia the first fully open parametric encyclopedia -- bridging factuality evaluation and knowledge materialization. All data, code, and a browsable interface are at https://llmpedia.net.