Understanding AI Credits in ArminCX
To provide AI features, ArminCX uses Large Language Models (LLMs). Each time the AI performs an action, AI credits are consumed. This article explains how credits are calculated, when they are consumed, and how to estimate usage.
1. What are AI Credits?
AI credits are the unit of measurement for AI feature usage. Instead of a flat fee for unpredictable computing loads, credits ensure fair, usage-based billing.
Since January 2026, variable AI costs are charged in addition to the monthly base fee.
2. When are Credits Consumed?
One credit is consumed per AI action. Actions fall into two categories:
A. Support Responses
Action | Credit Cost |
|---|---|
Auto-Response (AI replies automatically) | 1 credit |
Suggestion / Draft Reply (AI generates a draft) | 1 credit |
Text Modifier (tone change, shorten, formalize) | 1 credit |
B. Workflow Automation
Action | Credit Cost |
|---|---|
AI Condition Node (AI evaluates logic) | 1 credit |
Extract Data Node (AI extracts fields from a message) | 1 credit |
Workflow Response (AI sends a message in a workflow) | 1 credit |
Static condition / IF-THEN branch / delay / API call | 0 credits |
3. Credit Consumption in Practice
Rule of thumb: every AI node = 1 credit. Static logic, delays, and API calls do not consume credits.
Example A: Extract Data with multiple fields
An "Extract Data" node that extracts order number, basket value, and customer tag costs 1 credit — not one per field. The entire node is one AI action regardless of how many fields are configured.
Example B: Workflow with AI Condition + AI Response
Step | Type | Credits |
|---|---|---|
AI Condition: "Is this a complaint?" | AI node | 1 |
Branch: Yes → route to agent | Static logic | 0 |
Branch: No → AI sends reply | AI node | 1 |
Total | 2 credits |
Example C: Full automation flow
Step | Type | Credits |
|---|---|---|
Extract Data: order number + delivery date | AI node | 1 |
AI Condition: order frequency = high? | AI node | 1 |
Static filter: country = DE | Static logic | 0 |
Workflow Response: AI sends personalized message | AI node | 1 |
Total | 3 credits |
4. Efficiency & Predictability via AI Agent V5
The AI Agent V5 introduced Supermemory, which keeps credit consumption stable and predictable:
Predictability: Credit usage per interaction stays consistent, even for complex knowledge bases.
Intelligent Knowledge Usage: Supermemory uses pre-extracted facts instead of searching through large datasets on every request, preserving your credit balance and improving response speed.
5. Monitoring Your Usage
Go to Settings.
Click Billing.
View your current AI credits used and variable usage units for the active period.
→ See also: Billing Help Center Article