HINOW model
hinow/hiembed
HiEmbed
Turns text into a vector so you can search by meaning, not by exact word.
Best for: Semantic search, RAG, recommendation, and deduplication
An embedding does not answer questions and does not write text — it measures closeness in meaning. Every use starts there.
- Semantic search — find the right answer even when the user writes it in other words.
- RAG — pick the passages that go into the context before the language model answers.
- Recommendation — "similar to this one" across products, articles, job posts, or tickets.
- Clustering and triage — group messages about the same subject without a hand-written list of categories.
- Duplicates — find the ticket already open, the repeated record, the question that already has an answer.
- 1
Embed the content
Split the documents into passages and call
/v1/embeddingsonce per passage. Store the vector alongside the text. - 2
Embed the question
At search time, turn the user's question into a vector with the same call.
- 3
Compare and rank
The vectors come back normalized, so the dot product is the cosine similarity itself — the higher it is, the closer the meaning.
curl https://api.hinow.ai/v1/embeddings \
-H "Authorization: Bearer $HINOW_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"hinow/hiembed","input":"Como redefinir minha senha?"}'The response carries the vector in data[0].embedding and the usage in usage.prompt_tokens. You can send a list in input and get several vectors in a single call — much faster than one call per text.
When to compare another option
If what you need is to generate or summarize text rather than measure closeness, the path is Chat Completions with one of the language models.
See the full guide
Parameters, batch size, token limit, cost, and building a semantic search end to end.