Skip to main content
Embeddings Refinery
curl --request POST \
  --url https://api.example.com/v1/refine/embeddings
{
  "text": "<string>",
  "start_index": 123,
  "end_index": 123,
  "token_count": 123,
  "embedding": [
    {}
  ]
}
The Embeddings Refinery adds vector embeddings to your chunks, enabling semantic search and retrieval.

​
Request

​
Parameters

​
chunks
array
required
Array of chunk objects to add embeddings to. Each chunk must have text, start_index, end_index, and token_count fields.
​
embedding_model
string
default:"minishlab/potion-retrieval-32M"
The embedding model to use. Supports Hugging Face and OpenAI models.
Need another model? Reach out to us at [email protected]

​
Response

​
Returns

Array of chunks with added embedding field.
​
text
string
The original chunk text.
​
start_index
integer
Starting position in original text.
​
end_index
integer
Ending position in original text.
​
token_count
integer
Number of tokens in the chunk.
​
embedding
array
Vector embedding as array of floats.

​
Examples

from chonkie.cloud import TokenChunker, EmbeddingsRefinery

chunker = TokenChunker(chunk_size=512)
chunks = chunker.chunk("Your text here...")

refinery = EmbeddingsRefinery(
    embedding_model="minishlab/potion-retrieval-32M"
)
refined_chunks = refinery.refine(chunks)