Implementation:Openai Openai python Embedding Response Model

Knowledge Sources	openai-python OpenAI Embeddings API
Domains	NLP, Embeddings, Data_Processing
Last Updated	2026-02-15 00:00 GMT

Overview

Concrete typed response models for extracting embedding vectors and usage data provided by the OpenAI Python SDK.

Description

CreateEmbeddingResponse contains a list of Embedding objects, each with a float vector and position index. The Usage object tracks token consumption. When numpy is available and base64 encoding is used, vectors are automatically decoded to numpy arrays.

Usage

Access vectors via response.data[i].embedding. Use response.data[i].index to match embeddings to input order. Check response.usage for token consumption.

Code Reference

Source Location

• Repository: openai-python
• File: src/openai/types/create_embedding_response.py
• Lines: L1-33
• File: src/openai/types/embedding.py
• Lines: L1-25

Signature

class CreateEmbeddingResponse(BaseModel):
    data: List[Embedding]
    model: str
    object: Literal["list"]
    usage: Usage

    class Usage(BaseModel):
        prompt_tokens: int
        total_tokens: int

class Embedding(BaseModel):
    embedding: List[float]
    index: int
    object: Literal["embedding"]

Import

from openai.types import CreateEmbeddingResponse, Embedding

I/O Contract

Inputs

Name	Type	Required	Description
response	CreateEmbeddingResponse	Yes	API response from embeddings.create()

Outputs

Name	Type	Description
data[i].embedding	list[float]	Float vector of model output dimensions
data[i].index	int	Position matching input order
model	str	Model used for embedding
usage.prompt_tokens	int	Input tokens consumed
usage.total_tokens	int	Total tokens consumed

Usage Examples

Extract Vectors

from openai import OpenAI

client = OpenAI()
response = client.embeddings.create(
    input=["Hello world", "Goodbye world"],
    model="text-embedding-3-small",
)

# Extract vectors in input order
vectors = [item.embedding for item in sorted(response.data, key=lambda x: x.index)]
print(f"Got {len(vectors)} vectors of dimension {len(vectors[0])}")
print(f"Tokens used: {response.usage.total_tokens}")

Compute Similarity

import numpy as np

def cosine_similarity(a, b):
    a, b = np.array(a), np.array(b)
    return np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))

response = client.embeddings.create(
    input=["I love programming", "Coding is my passion", "The weather is nice"],
    model="text-embedding-3-small",
)
vecs = [d.embedding for d in response.data]
print(f"Similar: {cosine_similarity(vecs[0], vecs[1]):.3f}")
print(f"Different: {cosine_similarity(vecs[0], vecs[2]):.3f}")

Related Pages

Implements Principle

• Principle:Openai_Openai_python_Embedding_Result_Processing

Requires Environment

• Environment:Openai_Openai_python_Python_3_9_Plus