Thenlper: GTE-Base

Thenlper: GTE-Base is an embedding model from Thenlper, listed on api14 under the model id thenlper/gte-base. It sits in the embeddings category, so it is not a chat or text generation model. Its job is to turn text into vectors that represent meaning, and the provider describes it as encoding English sentences and paragraphs into a 768-dimensional dense vector space. It was released by the provider on 2025-11-18.

The model takes text as its only input and returns embeddings as its only output. Its context length is 512 tokens, and the maximum output is 460. Within that window it handles sentences and paragraphs, which matches the provider's description of the model as optimized for textual similarity, semantic search, and clustering. Because the output is a vector rather than generated text, you should expect numeric embedding data back from each call, not a written answer.

Those capabilities point to a fairly specific set of uses. Semantic search is the obvious one: embed your documents once, embed the query at request time, and rank by vector similarity. Clustering and topic grouping follow from the same property, since texts that are close in meaning end up close in the vector space. Textual similarity also covers deduplication, near-duplicate detection, and building similarity scores between two pieces of English text. Retrieval pipelines that feed a separate generation model are another natural fit, with GTE-Base handling the retrieval half.

You call it through api14's OpenAI-compatible API using the model id thenlper/gte-base. Because the endpoint follows the OpenAI format, you can point an existing OpenAI client at api14 and send your text as input in the usual way. Note that this model lists no supported parameters, so there are no extra sampling or generation options to set on the request. Keep your inputs within the 512 token context limit, and treat the response as embedding vectors rather than message content.

The main constraint to plan around is the 512 token context window. It suits sentences and paragraphs, but long documents will need to be split into chunks before embedding, with the chunking strategy chosen to preserve meaning across boundaries. If your task is text generation, summarisation, or anything conversational, an embeddings model is the wrong tool and a chat model will fit better. GTE-Base is aimed at English text, so other languages are not covered by the provider description.