Credit Cost per Model
Two things decide how many credits a task uses: which model runs it and how heavy the task is. This page gives you the rough cost of each, so you can predict your spend without doing any math.
Cost by model
Models fall into two tiers. Budget models handle the everyday work; premium models step in when a task genuinely needs more power.
| Tier | Models | Relative cost |
|---|---|---|
| Budget | Qwen, Groq | Lowest |
| Standard | OpenAI | Mid |
| Premium | Claude, Gemini | Highest |
A premium model can cost several times more than a budget model for the same amount of work. PrettiFlow routes light steps to budget models automatically, so most of your spend lands only on the steps that truly need a premium model.
Cost by task
Heavier tasks draw more credits, regardless of model. From lightest to heaviest:
| Task | Relative cost |
|---|---|
| Understanding your prompt | Lowest |
| Suggestions | Low |
| Planning | Low |
| Chat | Medium |
| Code generation | High |
| Tool calls | High |
| Deep reasoning | Highest |
What this looks like in practice
Light & cheap
Heavy & premium
A note on credits
Credit usage is recorded for every build, so you always have a full audit trail of where your credits went. For how credits map to plans and top-ups, see Credits explained.