For each action (e.g., text request, image generation, or speech recognition) a certain number of credits is deducted:
Text-to-Text:
Credit consumption depends on the chosen AI model, the amount of processed text (input and output), and – for reasoning models – additionally on the complexity of the request.
As a guideline: An average request with the ChatGPT 4.1 model corresponds to approximately 1 credit. Actual consumption may vary depending on length, scope, and model.
In the Language Model Hub you can see the exact credit prices per model and how many credits are charged per 10,000 characters (10,000 characters correspond to approximately 4 A4 pages in normal font size).
Speech-to-Text:
For speech-to-text conversion, a flat rate of 3 credits per action is deducted.
Image-to-Text:
For extracting information from images, a flat rate of 4 credits per action is deducted.
Text-to-Image:
Credit consumption depends on the chosen model, image size, and quality level:
GPT Image 1
- Low, 1024×1024: 1 Credit
- Low, 1024×1526 or 1526×1024: 1.5 Credits
- Medium, 1024×1024: 3.9 Credits
- Medium, 1024×1526 or 1526×1024: 5.8 Credits
- High, 1024×1024: 15.4 Credits
- High, 1024×1526 or 1526×1024: 23 Credits
FLUX 1.1 Pro
You can view your personal credit consumption and remaining balance at any time on the Subscription page.
Note: All information is for guidance purposes. Actual credit prices may vary and are updated regularly.