Sentence Transformers: paraphrase-MiniLM-L6-v2

Sentence Transformers: paraphrase-MiniLM-L6-v2 is an embedding model from Sentence Transformers, listed on api14 under the embeddings category. It is a compact MiniLM-based model whose purpose is to turn text into vectors rather than to generate text. The provider release date recorded for it is 2025-11-18.

The model accepts text as input and returns embeddings as output. It converts sentences and short paragraphs into a 384-dimensional dense vector space, producing semantic embeddings optimized for paraphrase detection, semantic similarity scoring, clustering, and lightweight retrieval. Its context length is 512 tokens, and the maximum output is 460. There are no other input or output modalities: it does not take images, audio or files, and it does not produce chat completions or free-form text.

Those capabilities point to a fairly specific set of uses. Because the vectors are trained to place paraphrases close together, the model fits semantic similarity scoring, where you compare two pieces of text and need a number that reflects meaning rather than word overlap. It suits paraphrase detection, clustering of sentences or short paragraphs into groups by topic, and lightweight retrieval, where you embed a query and a set of candidate passages and rank the passages by distance. Deduplication of near-identical entries, grouping support tickets or feedback, and building a small semantic search index over short documents are all natural fits. The 512 token context means each input should be a sentence or a short paragraph rather than a long document; longer material needs to be split into chunks before embedding.

You call it through api14's OpenAI-compatible API using the model id "sentence-transformers/paraphrase-minilm-l6-v2". Because the endpoint follows the OpenAI embeddings shape, you send your text as input and read the vector back from the response. The model lists no supported parameters, so there are no extra sampling or generation options to set; keep the request to the model id and the input text. The live price table on this page covers any billing details.

The main constraint to plan around is the 512 token context. If your inputs are long documents, or if you need generated text rather than vectors, this model is not the right choice and a text generation model would fit better. For sentence-level and short-paragraph embedding work, it is a small and direct option.