Docs
  1. Python Samples
Docs
  • Introduction
  • Quick Start Guide
  • Make a request
  • Chat Models
    • ChatGpt
      • ChatGPT (Audio)
        • Create a voice
        • Create a transcript
        • Create translation
      • ChatGPT (Chat)
        • Chat completion object
        • Create chat completion (streaming)
        • Create chat completion (non-streaming)
        • Create chat image recognition (streaming)
        • Create chat image recognition (streaming) base64
        • Create chat image recognition (non-streaming)
        • Function calling
        • N choices
        • Create chat function call (only non-streaming)
        • Create structured output
      • ChatGPT (Completions)
        • Completion object
        • Creation completed
      • ChatGPT(Embeddings)
        • Embedded Object
        • Create embed
    • Anthropic Claude
      • Create chat completion (streaming)
      • Create chat completion (non-streaming)
      • Create chat image recognition (streaming)
      • Create chat image recognition (non-streaming)
    • Gemini
      • Gemini Image creation interface (gemini-2.0-flash-exp-image-generation)
      • Chat interface
      • Image recognition interface
  • Image Models
    • MJ
      • Submit Imagine task (mj_imagine)
      • Submit Blend task (mj_blend)
      • Submit Describe task (mj_describe)
      • Submit Change task (mj_variation, mj_upscale,mj_reroll)
      • Query task status based on task ID
    • Ideogram
      • Generate with Ideogram 3.0
      • Edit with Ideogram 3.0
      • Remix with Ideogram 3.0
    • Kling Image
      • Submit Image Generation
      • Get Image by Task ID
      • Submit Kolors Virtual Try On
      • Get Kolors Virtual Try On by Task ID
    • DALL·E 3
      POST
    • Flux (OpenAI dall-e-3 format)
      POST
  • Video Models
    • Kling Video
      • Create Video by Text
      • Get Video by Task ID(text2video)
      • Create Video by Image
      • Get Video by Task ID(image2video)
    • Runway ML Video
      • Create Video by Runway
      • Get Video by Task ID
    • Luma Video
      • Create Video by Luma
      • Get Video by Task ID
    • Pika Video
      • Create Video by Pika
      • Get Video by Task ID
  • Music Model - Suno
    • Illustrate
    • Parameter
    • Task submission
      • Generate songs (inspiration, customization, continuation)
      • Generate lyrics
    • Query interface
      • Query a single task
  • Python Samples
    • python openai official library (using AutoGPT, langchain, etc.)
    • Python uses speech to text
    • Python uses text to speech
    • Python uses Embeddings
    • python calls DALL·E
    • python simple call openai function-calling demo
    • python langchain
    • python llama_index
    • Python uses gpt-4o to identify pictures-local pictures
    • python library streaming output
    • Python uses gpt-4o to identify images
  • Plug-in/software usage tutorials
    • Setting HTTP for Make.com with Yescale
    • Sample Code for gpt-4o-audio/gpt-4o-mini-audio
  • Help Center
    • HTTP status codes
  1. Python Samples

Python uses Embeddings

Embeddings API Guide#

Overview#

New Model Releases#

text-embedding-3-small
text-embedding-3-large
Features: Lower cost, improved multilingual performance, customizable dimensions.

Key Use Cases#

🔍 Search (relevance ranking)
📊 Clustering (similarity grouping)
👍 Recommendation Systems
⚠️ Anomaly Detection
📈 Diversity Analysis
🏷️ Text Classification

Basic Usage#

Retrieve Embedding Vectors#


Response Format#

{
  "object": "list",
  "data": [
    {
      "object": "embedding",
      "index": 0,
      "embedding": [
        -0.006929283495992422,
        -0.005336422007530928,
        // ... more values
      ]
    }
  ],
  "model": "text-embedding-3-small",
  "usage": {
    "prompt_tokens": 5,
    "total_tokens": 5
  }
}

Model Comparison#

ModelPages per DollarMTEB PerformanceMax Input Tokens
text-embedding-3-small62,50062.3%8191
text-embedding-3-large9,61564.6%8191
text-embedding-ada-00212,50061.0%8191

Practical Applications#

Process Review Data#


Technical Details#

Dimensions#

text-embedding-3-small: Default 1536 dimensions
text-embedding-3-large: Default 3072 dimensions
Dimensions can be adjusted via the Dimensions parameter.

Notes#

Billing is based on the number of input tokens.
Approximately 800 tokens per page.
All models support a maximum input of 8191 tokens.
Previous
Python uses text to speech
Next
python calls DALL·E
Built with