Glossary · Term

text-embedding-three-large

← all terms

Definition

Plain language

A tool from OpenAI that turns a passage of text into a list of numbers capturing its meaning, so a computer can measure how similar two passages are.

As stated in the literature

OpenAI's large text-embedding model, mapping text into a high-dimensional vector space where cosine distance approximates semantic similarity; used in the semantic-collapse study as the measurement substrate for tracking meaning drift across multi-LLM conversations.

Also called: text-embedding-3-large

Why it matters: By converting text into comparable numbers, it lets software measure how alike two passages are and track how meaning drifts across a long conversation.

For example, it can turn 'the dog chased the ball' and 'a puppy ran after a toy' into number lists that come out close together because the meanings are similar.

Heard on the show

“Every claim about semantic diversity rests on cosine distances measured in a particular embedding space — OpenAI's text-embedding-three-large.”
Episode 073 — When Three LLMs Talk to Each Other, Their Ideas Quietly Stop Moving

Mentioned in 1 episode

  1. 073
    When Three LLMs Talk to Each Other, Their Ideas Quietly Stop Moving

Related terms