Skip to main content
OpenAI’s pricing is dynamic and may change at any time. For the most current rates, please visit openai.com/pricing.
OpenAI categorizes its foundation models into three primary types—text, image, and audio—each charged using a distinct unit:
Model CategoryBilling UnitBasis
Text1,000 tokens~750 words per 1,000 tokens
ImagePer generated imageCost varies by image resolution
Audio (Whisper)Per second of audioCharged by duration (transcription/translation)
A token corresponds to roughly 0.75 English words. Tools like tiktoken help estimate token counts before you send requests.

​
Text-Based Use Cases

Text models support a variety of tasks, each billed per 1,000 tokens:
Use CaseDescription
Word/Code CompletionPredictive text or code generation
ChatConversational AI interactions
EmbeddingsSemantic vector generation
Fine-TuningCustom model training
The image shows the OpenAI pricing model for text, image, and audio services, detailing costs per 1,000 tokens for various tasks like word/code completion, chat, embedding, fine-tuning, and Whisper audio.
Always check the official pricing page before planning your integration or budgeting for production workloads.
We’ll now take a closer look at ChatGPT—how it works and how to integrate it into your applications.

Watch Video