嵌入

POST /v1/embeddings 會透過 CometAPI 使用所選模型產生文字嵌入，用於語意搜尋、分群與擷取工作流程。

POST /v1/embeddings

Python

Single Text

from openai import OpenAI

client = OpenAI(
    base_url="https://api.cometapi.com/v1",
    api_key="<COMETAPI_KEY>",
)

response = client.embeddings.create(
    model="text-embedding-3-small",
    input="The food was delicious and the waiter was friendly.",
)

print(response.data[0].embedding[:5])  # First 5 dimensions
print(f"Dimensions: {len(response.data[0].embedding)}")

200

{
  "object": "list",
  "data": [
    {
      "object": "embedding",
      "index": 0,
      "embedding": [
        -0.0021,
        -0.0491,
        0.0209,
        0.0314,
        -0.0453
      ]
    }
  ],
  "model": "text-embedding-3-small",
  "usage": {
    "prompt_tokens": 2,
    "total_tokens": 2
  }
}

​

概覽

嵌入 API 會產生可捕捉語意的文字向量表示。這些向量可用於語意搜尋、分群、分類、異常偵測，以及擷取增強生成（RAG）。

CometAPI 支援來自多個供應商的嵌入模型。傳入一個或多個文字字串後，會回傳數值向量，您可以將其儲存在向量資料庫中，或直接用於相似度計算。

​

可用模型

Model	Dimensions	Max Tokens	Best For
text-embedding-3-large	3,072（可調整）	8,191	最高品質的嵌入
text-embedding-3-small	1,536（可調整）	8,191	具成本效益、速度快
text-embedding-ada-002	1,536（固定）	8,191	舊版相容性

請參閱模型清單以查看所有可用的嵌入模型與定價。

​

重要說明

降低維度 — text-embedding-3-* 模型支援 dimensions 參數，讓您可以在幾乎不明顯影響準確度的情況下縮短嵌入向量。這可將儲存成本最多降低 75%，同時保留大部分語意資訊。

批次輸入 — 您可以透過將字串陣列傳入 input 參數，在單一請求中為多段文字建立嵌入。與針對每段文字分別發送請求相比，這樣的效率高得多。

---

授權

Authorization · string · header · 必填

Bearer token authentication. Use your CometAPI key.

主體

application/json

model · string · 必填

The embedding model to use. See the Models page for current embedding model IDs.

範例:

"text-embedding-3-small"

input · string string[] integer[][] · 必填

The text to embed. Can be a single string, an array of strings, or an array of token arrays. Each input must not exceed the model's maximum token limit (8,191 tokens for text-embedding-3-* models).

encoding_format · enum<string> · 預設值: float

The format of the returned embedding vectors. float returns an array of floating-point numbers. base64 returns a base64-encoded string representation, which can reduce response size for large batches.

可用選項: float, base64

dimensions · integer

The number of dimensions for the output embedding vector. Only supported by text-embedding-3-* models. Reducing dimensions can lower storage costs while maintaining most of the embedding's utility.

必填範圍: x >= 1

user · string

A unique identifier for your end-user, which can help monitor and detect abuse.

回應

200 - application/json

A list of embedding vectors for the input text(s).

object · enum<string>

The object type, always list.

可用選項: list

範例:

"list"

data · object[]

An array of embedding objects, one per input text. When multiple inputs are provided, results are returned in the same order as the input.

Show child attributes

model · string

The model used to generate the embeddings.

範例:

"text-embedding-3-small"

usage · object

Token usage statistics for this request.

Show child attributes