LLMpedia is an innovative research project dedicated to exploring the capabilities of large language models (LLMs) in generating knowledge-based content. This extensive encyclopedia houses over one million AI-generated articles, created entirely from the memory of its model, GPT-5-mini. The platform focuses on analyzing the factuality and parametric knowledge of LLMs, shedding light on what these models have learned from their extensive training data.

Users can browse through a vast array of topics, ranging from historical events and notable figures to scientific concepts and cultural phenomena. In addition to article browsing, LLMpedia offers functionalities for downloading datasets and evaluating various configurations. The initiative is designed not just to provide information but also to encourage further research into the accuracy and biases of AI-generated content.

LLMpedia represents a significant step in understanding the influence of artificial intelligence on knowledge dissemination and access, paving the way for future advancements in the field