OpenAI: Text Embedding 3 Large

OpenAI: Text Embedding 3 Large is an embedding model from OpenAI, listed on api14 under the model id openai/text-embedding-3-large. It sits in the embeddings category rather than the chat category, which means it is built to turn text into vectors rather than to generate conversational replies. OpenAI describes text-embedding-3-large as its most capable embedding model for both English and non-English tasks.

The model takes text as its only input and returns embeddings as its output: numerical representations of text that can be used to measure the relatedness between two pieces of text. It supports a context length of 8192 tokens, and its maximum output is 7372. No additional request parameters are supported for this model, so there are no tuning knobs beyond what the embedding endpoint itself accepts.

Because the output is a vector rather than prose, the practical uses follow from comparing those vectors. Typical fits include semantic search, where you embed a query and a corpus and rank documents by similarity; retrieval for a RAG pipeline, where embeddings decide which passages to feed a chat model; clustering or deduplicating large text collections; classification and routing by nearest example; and recommendation or relatedness scoring between two passages. The provider's note that the model handles both English and non-English tasks makes it a reasonable choice for multilingual corpora where you want one embedding space across languages.

To call it through api14, use the OpenAI-compatible API with the model id openai/text-embedding-3-large. Send your text to the embeddings endpoint in the same way you would against OpenAI directly, and read the vectors back from the response. Because supported_parameters is empty, you should not expect to pass extra options such as dimensions or encoding format through this gateway; keep requests to the text you want embedded.

The main limits to keep in mind are the input modality and the context length. Only text goes in, so images, audio and other formats are out of scope for this model. Any single input must fit within 8192 tokens, which means long documents need to be split before embedding. And since it produces embeddings rather than text, it cannot answer questions or write content on its own; pair it with a chat model when you need generation on top of retrieval.