📝 Overview
In AI worker, the total number of task executions is calculated by summing three types of activity.
AI usage
Tool usage
AI worker usage in Flowbot
1️⃣ AI Usage
Tasks are consumed based on the number of tokens used by each AI model.
Tokens Per Tasks (By Model)
Model | Input tokens | Output tokens |
GPT 5-mini | 20,000 | 2,500 |
GPT 5.1 | 4,000 | 500 |
GPT 5.2 | 2,800 | 300 |
Gemini 2.5 Flash | 16,600 | 2,000 |
Gemini 3 Flash | 10,000 | 1,600 |
Gemini 3 Pro | 1,200 | 200 |
Gemini 2.5 Pro | 4,000 | 500 |
Claude Haiku 4.5 | 5,000 | 1,000 |
Claude Sonnet 4.5 | 1,600 | 300 |
📍 Example:
If you use 20,000 input tokens and 8,000 output tokens with Gemini 3 Flash, you will consume 7 tasks (2 input tasks + 5 output tasks).
2️⃣ Tool Usage Tasks
Tasks will be consumed when the AI worker works with other tools, such as performing searches or creating documents.
Tool | Tasks consumed | Notes |
Operating apps | 1 | 1 task per action executed. |
Creating documents | 1 | 1 task per document created. |
Searching | 5 | 5 tasks per search performed. |
3️⃣ AI Workers Usage in Flowbot
When you run an AI worker as part of a Flowbot, 1 task is always consumed per run. This applies only when the AI worker is used within a Flowbot.
⚠️ Important Notes
Test runs: Tasks are used during test runs as well.
Success/Failure: Tasks are used regardless of whether the AI worker processing succeeds or fails. Make sure your prompts are clear!
📚 Related Guides
What is an AI Worker? - View Guide Here
AI Worker | Basic Setup Guide - View Guide Here
AI Worker | How to Create a Guide - View Guide Here
Search Keywords
Search Keywords
AI worker, task, task usage, token count, cost management, GPT, Gemini, Claude, test run, usage