Predictions
Understand how predictions work on Skytells and how they are billed.
What Are Predictions?
Predictions are the core functionality of the Skytells API — they allow you to generate content using various AI models. With the prediction endpoints, you can create images, text, videos, or combinations of content types by providing a prompt and model-specific parameters.
What You Can Generate
Text
Generate natural language responses, summaries, translations, and structured outputs using language models.
Images
Create images from text prompts — product visuals, illustrations, concept art, and more.
Video
Generate video content from prompts or input media using video generation models.
Multi-Modal
Combine text, image, and video inputs and outputs for complex generation workflows.
How Predictions Are Billed
The billing method for predictions depends on the model you use. Each model has its own pricing criteria, which you can find on the model's page in Dashboard → Console.
| Model Type | Typical Billing Method |
|---|---|
| Language Models | Per token (input + output tokens) |
| Image Generation | Per prediction or per megapixel |
| Video Generation | Per prediction or per second of output |
| GPU-Accelerated Models | Per second of compute time |
Some models combine billing methods — for example, a model may charge per prediction with additional per-second GPU costs for longer processing times.
Finding Model Pricing
Every model on Skytells displays its pricing and billing method on its page. You can find this information from:
- Dashboard → Console — select a model to view its pricing
- Skytells Models — flagship model pricing
- Explore Models — third-party model pricing
Cost Tips
Need Help?
Contact Support
Submit a support ticket for prediction pricing or billing questions.
How is this guide?