Name: Text Embedding Ada 002 API
Brand: OpenAI

Overview

text-embedding-ada-002 Description

Basic Information

Model Name: text-embedding-ada-002
Developer/Creator: OpenAI
Release Date: December 2022
Version: text-embedding-ada-002
Model Type: Text Embedding

Overview

Text-embedding-ada-002 is an efficient and reliable embedding model designed to convert text into numerical representations. It serves as a foundational tool for various natural language processing (NLP) applications, enabling machines to understand and process human language more effectively.

Key Features

High Dimensionality: Provides embeddings with 1536 dimensions, capturing detailed semantic information.
Broad Applicability: Suitable for a wide range of NLP tasks, including search, clustering, and classification.
Scalability: Optimized for handling large datasets and high-volume requests, making it ideal for enterprise applications.

Intended Use

Search: Enhances search engines by ranking results based on relevance to the query.
Clustering: Groups similar text strings together, useful in organizing large datasets.
Recommendations: Improves recommendation systems by identifying related items.
Anomaly Detection: Identifies outliers in datasets, which can be critical for security and quality control.
Diversity Measurement: Analyzes similarity distributions to ensure diverse content representation.
Classification: Assigns text strings to predefined categories based on similarity.

Text-embedding-ada-002 can also be used for Medical Coding. Model successfully identifies the relevant code from a set of similar codes 80% of the time (better than GPT 4 with 50%). Learn more about this and other models and their applications in Healthcare here.

Technical Details

Architecture:
- Utilizes a Transformer-based architecture known for its efficiency in processing sequential data. Transformers excel in capturing contextual relationships between words in a sentence, leading to better semantic understanding.
Training Data:
- Trained on a diverse and extensive dataset sourced from various internet texts, including books, articles, and web pages. This diverse training data helps the model generalize well across different domains and applications.
Data Source and Size:
- Leveraged a vast corpus of text data, ensuring comprehensive coverage of language use cases. The large-scale training dataset allows the model to capture nuanced language patterns.
Knowledge Cutoff:
- The model has a knowledge cutoff of September 2021, meaning it was trained on data available up to this date. It does not include information or events occurring after this period.
Diversity and Bias:
- Efforts were made to include a diverse range of text sources to minimize biases. However, some biases may still exist due to the nature of the training data. Continuous evaluation and updates are necessary to address any identified biases.

Performance Metrics

Comparison to Other Models:
- Outperformed many predecessors and comparable models at the time of its release, especially in terms of cost-efficiency and scalability.
Accuracy:
- Demonstrated strong performance on key benchmarks:
  - MIRACL: Achieved an average score of 31.4%, reflecting its capability in multi-language retrieval tasks.
  - MTEB: Scored 61.0% on average, indicating solid performance in English language tasks.
Speed:
- Optimized for quick inference, making it suitable for real-time applications and services.
Robustness:
- Capable of handling a variety of input types and maintaining performance across different text formats and languages.

Uptime and performance

Uptime (7d): 20.0%

Measured on the AI/ML API gateway across live traffic. Figures are medians unless stated. Last updated 2026-10-12.

Use cases

Semantic search & retrieval

Use Ada‑002 for embedding queries and documents to power content-based search in knowledge bases and CMS platforms.

Document classification & clustering

Apply embeddings to categorize content, tag documents, or create topic clusters semantically.

Recommendation systems

Leverage embedding similarity to suggest related articles, products, or media items.

RAG workflows

Combine dense embeddings with vector databases to retrieve context for downstream generation or chatbot tasks.

Anomaly detection

Identify unusual or off-topic content by measuring embedding divergence in streams or logs.

‍

How to use Text Embedding Ada 002 API

Install any OpenAI-compatible SDK, point it at api.aimlapi.com/v1, and set the model to openai/text-embedding-ada-002.

import requests

r = requests.post(
    "https://api.aimlapi.com/v1/embeddings",
    headers={"Authorization": "Bearer " + AIMLAPI_KEY},
    json={
      "model": "openai/text-embedding-ada-002",
      "input": "The quick brown fox"
    },
)
print(r.json())

const r = await fetch("https://api.aimlapi.com/v1/embeddings", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.AIMLAPI_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    "model": "openai/text-embedding-ada-002",
    "input": "The quick brown fox"
  }),
});
console.log(await r.json());

curl -X POST https://api.aimlapi.com/v1/embeddings \
  -H "Authorization: Bearer $AIMLAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/text-embedding-ada-002","input":"The quick brown fox"}'

OpenAI-compatible — swap the base URL and it works with your existing SDK.