RelayX

Embeddings

Create dense vectors for semantic search, RAG indexing, clustering, and deduplication. RelayX currently routes Chinese-first embeddings to DashScope text-embedding-v4.

POST /v1/embeddings

Try it

curl https://relayx.timor419.com/v1/embeddings \
  -H "Authorization: Bearer $RELAYX_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "dashscope/text-embedding-v4",
  "input": "RelayX 是一个 OpenAI 兼容 API 网关。",
  "dimensions": 1024
}'

Parameters

  • `input`: a string or array of strings. Batch inputs return one vector per item; text-embedding-v4 accepts up to 10 items per request.
  • `dimensions`: optional vector size. text-embedding-v4 defaults to 1024 and supports up to 2048.
  • Billing is based on input tokens; responses include X-RelayX-Usage with input_tokens and vectors.