Create Embeddings
Create embeddings through nRouter with one POST to /v1/embeddings, with models from OpenAI, Cohere, Voyage, and Vertex for search and RAG workloads.
Generate dense vector embeddings for semantic search, retrieval-augmented generation (RAG), and clustering via a single POST request to /v1/embeddings across industry-standard vector providers.
Request Syntax & Options
Generate embeddings across OpenAI, Cohere, Voyage, and Vertex AI models (live catalog):
curl https://api.nrouter.ai/v1/embeddings \
-H "Authorization: Bearer $NROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model": "text-embedding-3-small", "input": "How do fallbacks work?"}'input accepts a string or array of strings. Configure encoding_format ("float" or "base64" to compress payload transfer) and dimensions (on text-embedding-3).
Token usage reports under x-nr-input-tokens and x-nr-total-tokens headers, settling at list price in request logs. See Embeddings API, Generate Speech, and Errors & Rate Limits.
Frequently asked questions
Which providers can I embed with?
OpenAI, Cohere, Voyage, and Google Vertex AI; inspect the live catalog for current models.
Can I embed many texts at once?
Yes; pass an array of strings in input to receive a corresponding array of embedding vectors in data.
How do I keep large batch responses small?
Set encoding_format to base64 to reduce JSON payload transfer overhead on large batches.
Do I need a separate key for embeddings?
No; a single virtual key and credit balance covers text, embedding, and multimodal endpoints.