👤 For customers with Pipefy AI Agents configured
🔐 Available on all plans with AI Agents
🎯 To understand what changes with the new AI Models and how it affects credit consumption
Starting July 13, 2026, Pipefy AI Agents offer three AI Models, Lite, Standard, and Pro, each with a different credit consumption. This FAQ answers the most common questions about what changes, how much each Model costs, and what to do, or not do, from now on.
📖 You will find answers about:
About the change
What exactly changed in Pipefy AI Agents?
Each behavior configured in a Pipefy AI Agent can now use one of three AI Models, Lite, Standard, or Pro. Before, only one Model was available, now called Classic. The Model is chosen per behavior, not per Agent, so a single Agent can combine different Models depending on the complexity of each task.
Do I need to do anything right now?
No. If you don't change anything, your AI Agents keep working exactly as they are, using the Classic Model. No action is needed until the Classic Model is discontinued, on September 30, 2026.
Does this mean my AI Agents will cost more?
Credit consumption depends on the Model chosen for each behavior, not on your plan. The Standard Model has the same credit consumption as Classic, so the automatic migration doesn't change your current consumption. Consumption only changes if you actively choose the Pro Model outside the promotional period.
If you're not sure which Model to pick, start with Standard. It covers most day to day operational behaviors.
AI Models and credits
What's the difference between the Lite, Standard, and Pro Models?
Lite consumes 2 credits per run and fits short, predictable tasks with structured input and a single result, like translating a field or classifying a request. Standard consumes 2 credits per run and fits medium complexity tasks, like reading documents, conditionals across multiple fields, or several actions in sequence. Pro consumes 3 credits per run, with a promotional rate of 2 credits until October 8, 2026, and fits high stakes or complex analysis work, with conflicting data sources or judgment calls that require more nuance.
If Lite and Standard cost the same, what's the benefit of using Lite?
Speed. Lite is the fastest of the three Models, so the benefit of choosing it isn't about saving credits, it's about how quickly you get a response. It's a good fit for high volume, low complexity behaviors where delivery speed matters more than reasoning depth.
Are there actions that consume additional credits?
Yes. Besides the chosen Model's consumption, three specific actions add extra credits when used inside a behavior: document analysis adds 1 credit, calculations or analysis add 0.5 credit, and web search adds 0.5 credit. This additional charge applies to all three Models, with no exceptions.
Is there any situation where a behavior doesn't consume credits?
Yes, two of them. When the automation is active but the configured condition isn't met, and when the Agent receives more data than the supported limit, such as more than 30 attachments in a single run. If there's a partial error but part of the run completes, consumption happens normally.
Migration and timeline
What happens to the Classic Model?
Classic is the name given to the Model that already existed before the new Models launched. It keeps working normally until September 30, 2026, when it will be discontinued.
What happens to me when Classic is discontinued?
All behaviors configured with the Classic Model automatically migrate to the Standard Model, which has the same 2 credit consumption.
If you want to keep the same credit consumption level after September 30, 2026, you don't need to do anything. The migration to Standard is automatic and keeps the same credit cost as Classic.
Will the Pro Model get more expensive?
Yes, but only starting October 9, 2026. Until October 8, 2026, Pro consumes 2 credits per run, the same as Standard, to make it easier to test. Starting October 9, 2026, it consumes 3 credits.
Do these changes apply to Single Tenant customers?
The timeline in this FAQ applies to multi tenant customers. Single Tenant customers follow the specific update schedule of their own instance, so the dates may differ. If you're a Single Tenant customer, confirm the dates with your customer success team.
In practice
Can I use different Models within the same AI Agent?
Yes. Each AI Agent can have up to 5 behaviors configured, and each one chooses its own Model independently.
Where do I choose the AI Model for a behavior?
In the behavior configuration screen of the AI Agent, when defining that behavior's trigger and instructions.
Where do I track my AI credit consumption?
In your organization's Admin Panel.
How does credit consumption work if I connect my own AI model through BYOM?
Differently. Each Agent run and each Automation run consumes 1.5 credit for Agent Reasoning, Actions, and Outputs, plus the specific charges for agentic capabilities like document analysis, calculations, or web search.
