Embeddings — API Reference

curl -sS -X POST "https://api.dev.dodil.io/v1/embeddings" \
  -H "Authorization: Bearer $DODIL_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "jina-embeddings-v4",
    "input": ["how do vector databases work?", "ANN index trade-offs"],
    "encoding_format": "float",
    "normalized": true,
    "dimensions": 512,
    "task_type": "RETRIEVAL_QUERY"
  }'

rpc Embed(EmbedRequest) returns (EmbedResponse);
 
message EmbedRequest {
  string model = 1;
  google.protobuf.Value input = 2;               // string | [strings] | [content objects with image_url]
  string encoding_format = 3;                     // "float" | "base64"
  bool normalized = 4;                            // default true
  optional int32 dimensions = 5;                  // Matryoshka truncation (optional)
  EmbedTaskType task_type = 6;
}
 
enum EmbedTaskType {
  EMBED_TASK_TYPE_UNSPECIFIED = 0;
  RETRIEVAL_QUERY = 1;
  RETRIEVAL_DOCUMENT = 2;
  SEMANTIC_SIMILARITY = 3;
  CLASSIFICATION = 4;
  CLUSTERING = 5;
  QUESTION_ANSWERING = 6;
  FACT_VERIFICATION = 7;
  CODE_RETRIEVAL_QUERY = 8;
  CODE_RETRIEVAL_DOCUMENT = 9;
}

grpcurl \
  -H "Authorization: Bearer $DODIL_TOKEN" \
  -d '{
    "model": "jina-embeddings-v4",
    "input": ["how do vector databases work?", "ANN index trade-offs"],
    "encoding_format": "float",
    "normalized": true,
    "dimensions": 512,
    "task_type": "RETRIEVAL_QUERY"
  }' \
  $IGNITE_GRPC \
  dodil.ignite.v1.ModelService/Embed

{
  "object": "list",
  "model": "jina-embeddings-v4",
  "data": [
    { "object": "embedding", "index": 0, "embedding": [0.0123, -0.0456, 0.0789] },
    { "object": "embedding", "index": 1, "embedding": [0.0234, -0.0567, 0.0890] }
  ],
  "usage": { "prompt_tokens": 14, "total_tokens": 14 }
}

message EmbedResponse {
  string object = 1;                              // "list"
  string model = 2;
  repeated Embedding data = 3;
  EmbedUsage usage = 4;
}
 
message Embedding {
  string object = 1;                              // "embedding"
  int32 index = 2;
  repeated float embedding = 3;
}
 
message EmbedUsage {
  int32 prompt_tokens = 1;
  int32 total_tokens = 2;
}