Full definition
LLM Visibility is the probability that a large language model — GPT-4, Claude, Gemini, GLM-4, Llama — will name your brand unprompted when asked an open-ended question in your category. The measurement is conceptually distinct from AI Search Visibility: LLM Visibility captures the model's internal memorization (what was learned during training), while AI Search Visibility captures live retrieval (what the model finds when it searches the web). A brand can have high AI Search Visibility but low LLM Visibility — meaning the model finds you when it searches, but does not yet "think" of you spontaneously. The fix for low LLM Visibility is a persistent-training campaign: getting cited on enough high-authority third-party domains (Wikipedia, G2, Crunchbase, industry analyst reports, government registries) that the next model checkpoint cannot avoid learning your brand as a category exemplar. Harch Atelier measures LLM Visibility with a quarterly panel of 200 category queries run across GPT-4, Claude 3.5, Gemini 1.5, and GLM-4 in zero-retrieval mode (web search disabled), then tracks the unweighted average citation rate as the LLM Visibility score. Brands above 40 percent are category-defining; brands below 10 percent are functionally invisible to the model's spontaneous reasoning.