Overview
Taskade Genesis routes prompts, AI agent runs, and AI automation steps through frontier models. The picker shows the credit cost before you start. Published apps run for free. Visitors do not use your credits.
TL;DR: Credits cover building and AI work. Published Taskade Genesis apps run free for visitors. The model picker shows the current cost before each step. Taskade EVE remembers your selected model across conversations. Monthly credits reset each billing cycle. Purchased and earned Usage credits do not expire. Compare plans.
Choose a Model Type
Choose a model by the depth of reasoning you need. Cost rises with the depth of reasoning, not the brand name.
| Model type | What it is great at | Credit cost | Best for |
|---|---|---|---|
| Lightweight & fast | Fast, cheap responses | Cheapest | Quick iterations, simple text, high-volume agent calls |
| Balanced everyday | Balanced everyday reasoning and code | Mid | General reasoning, code, balanced quality |
| Open-source value | Low-cost open-weight models | Low to mid | Cheap routine steps, bulk routing, cost-sensitive automations |
| Long-context | Room for very long inputs | Higher | Long documents, deep analysis |
| Deep reasoning | Deepest step-by-step thinking | Highest | Math, multi-step planning, complex logic |
The model picker lists TSK-Auto first, then the TSK tiers in order of depth. TSK-Auto is the recommended default. Hover a tier to see its credit cost before you run it:
┌──────────────────────────────────────────────────────────┐
│ Taskade Genesis › Prompt bar › Model │
├──────────────────────────────────────────────────────────┤
│ ● TSK-Auto Recommended · picks the best model │
│ TSK-Instant Fast edits │
│ TSK-Standard Everyday tasks │
│ TSK-Thinking Multi-agent planning │
│ TSK-Reasoning Workflows and systems │
├──────────────────────────────────────────────────────────┤
│ Current Model › pin one model, cost on hover │
└──────────────────────────────────────────────────────────┘
Taskade adds new models as they ship. If you do not want to choose, use Auto mode. Auto picks the right tier for the job. You pay for depth only when you need it.
Taskade EVE remembers the model you select. Your next conversation starts with that choice. If a saved model is no longer available, Taskade EVE returns to Auto.
Compare Credit Pools
Compare Monthly credits and Usage credits as separate pools. Taskade uses Monthly credits first.
| Credit pool | How it works |
|---|---|
| Monthly credits | Included with an eligible plan and reset each billing cycle |
| Usage credits | Purchased or earned credits that do not expire |
| Earned credits | The Usage credit pool shown on Free |
See pricing for current plan allowances.
Reduce Credit Use
- Use lightweight models for cheap iterations and high-volume agent calls
- Reserve reasoning models for math, planning, and hard logic
- Publish your app as soon as it is stable. Published apps cost zero credits to run
- Let Auto mode route the prompt when you are unsure
Keep Published Apps Free to Run
Publish a stable app before you share it with visitors.
Build once, run forever. You spend credits when you edit an app in the builder. Visitors do not spend your credits on the published version. Any paid plan gives you credits to build and edit. Those apps then serve unlimited visitors.
Related guides
- AI Models. Choose from the current model catalog.
- Publishing. Publish to lock in the free-runtime behavior.
- Custom AI Agents. Choose a model for a specific agent.