Yeah what I meant is that the āuse your own connectionā is not available for core plans, only pro and up (see Credits - Help Center). Free and Core plans can only make use of Makeās AI providor which is as I know GPT-5-nano.
Typically when one starts to consider a tool to use one starts with free plan or an entry level plan like the Make core plan. So letās say somebody has an idea for an ai workflow or chatbot. They try out several solutions one of them being Make. Make has a nice workflow, good management tools but somehow the ai output isnāt as sophisticated as in other solutions. Make runs the risk that one disregards it because of this. And yes I know that one could use the http model in core to use the more advanced llm models, but the point is that new users encounter more friction in considering if make is a good tool for them.
Iām really not happy about this at all. I can already see that in the 2 organizations I manage, credit usage is going to be much higher with this new billing system. I saw that an automation that modifies 10 lines of a spreadsheet counts as 10 credits, and the same applies to the iterator. I still have to test the agents part more thoroughly, but I can imagine what it will be like. Itās like saying weāre going to raise prices, but instead of being direct about it, we call them credits to make it even more complicated to understand.
I understand your frustration, and I appreciate you taking the time to share this feedback. To clarify your question: in general, one operation equals one credit, but there are exceptions. Certain features, especially AI-related actions or custom code, may consume more than one credit per operation or calculate usage based on variable inputs.
Actually, it always worked like that. There is no change in this behaviour.
Every time your scenario is triggered, it consumes one credit, no matter how many data bundles (items) are processed/output by the ātrigger moduleā.
Every action done by subsequent modules (usually one action per one data bundle/item) costs one credit, no matter how many data bundles/items the specific module outputs.
The change impacted only the Makeās in-house AI modules.
Regarding Google Sheets - you can always use āBulk Update Rows (Advanced)ā module and update the whole sheet for 2 credits (Iterator + Bulk Update Rows (Advanced)).
Think about it: if the change makes 0 impact on the user - why changing the billing for the user then ?
Thatās simply a perception game to make it look insignificant - but which opens the door to a credit-based billing that can later on be adapted for higher operation rate cost per credit unit, without changing the credits since itās now masking the operations.
The way the community managers are answering are great tells: theyāre simply repeating the same over and over again āone op = one creditā yeah but till when ? Highly doubt itās gonna be forever - otherwise they would say it
Now that Iāve had a bit more time to go into more detail, I realize that in my most basic automations where before it was one operation, now itās x3, and in others that are a bit more elaborate, itās x20 or x30. This is going to have a strong impact on my operations.
Plus, @Jobless_Tobacconist commented that I have to use Bulk Update Rows (Advanced) module. Imagine the work we would have to do to change dozens of scenarios, and it would still be a x3.
They need to change the billing to accomodate the charging of their AI tools. Third party AI API costs need to be charged through. To be fair, I totally get that.
just to be clear - if iām using a Make module that accesses Google Gemini or ChatGPT or Claude or DeepSeek - there should be no additional costs from Make - right?
Youāre missing the whole point of what weāre saying here. This new model makes you run out of credit faster than before, forcing you into upgrading your plan earlier than you shouldāve
What Iāve noticed is that a scenario run that used to cost 7 operations, now costs 7 credits. So iām not seeing the point you are trying to make. If you donāt use the new Make AI tools, you wonāt run out of credits earlier. But please correct me if Iām missing somewhat.
This change is pure greed!
I only have one quite simple automation WITHOUT AI features.
It runed multiple times a day and its now a few hundret credits per day.
that means, to only run it ONCE a day Iād have to play $10 a monthā¦.
I was already working on other automations, but to keep working on them and even run them at least once a day Iād need to spend $120 /month, just to have the same volume as the $10/month had a few days agoā¦.
I love Make.com, but I simply cant afford couple hundrets a moth, just to have the same experience.
Hello,
Yesterday, August 27th, we attempted to run an automation but received a message stating that all credits had been used, even though the dashboard showed 4,368 credits remaining (see attachment). This suggests that either the message or the displayed number is incorrect.
We purchased an additional 1,000 credits and were able to perform one more run of the automation, but the same message appeared again. The dashboard now shows 5,108 credits remaining. I have already submitted a support ticket, but I am posting this here for visibility to the team. This issue is really blocking us, as purchasing more credits does not seem to be a solution.
Iām Petr, I work on special projects in our revenue team, and Iāve been directly responsible for leading the move to credits and defining how the calculations work. I wanted to jump in here personally, because Iāve read through your feedback and concerns and I think itās important you hear directly from me.
First, thank you for being open ā even when itās critical. I know a change like this impacts your day-to-day, and itās only fair that we explain it transparently.
Hereās the essence of whatās changing (and whatās not):
For 99% of modules, nothing has changed. What used to be 1 operation is now 1 credit. For example, iterating over 10 rows in a spreadsheet was always 10 operations ā now itās 10 credits. That logic is unchanged.
Where it is different is with AI modules using Makeās AI provider. Here, credits are based on token usage, because the underlying AI costs scale with input and output size. This is the only area where consumption may feel higher than before.
If you bring your own API key (OpenAI, Claude, Gemini, etc.), on plans starting from Pro and higher you can use it with supported AI features. In that case, itās always a flat 1 credit per run ā just like before ā and you pay token costs directly to your provider, not to Make.
Why we changed: the usage-based model (charged via credits) lets us keep simple modules affordable, while aligning advanced features with the resources they actually consume. The flat-rate model (charged via operations) didnāt reflect the true cost difference between a simple action (like sending an email) and a compute-heavy one (like summarizing a 10-page document). Credits make that distinction transparent and fair.
I also hear the perception: some of you are worried about scenarios reporting higher usage. In some cases, thatās a misunderstanding (because the same number of operations were already counted before), and in others it could be a bug. If you see scenarios where usage looks drastically higher than before, please share them ā weāll investigate together. Also, weād encourage you to submit a support ticket via https://www.make.com/en/ticket so we keep track of all credit-related bugs.
I know this is a big change, and it will take time to adjust. But my goal here is transparency: to show you exactly how it works, to acknowledge where itās different, and to give you options on how you want to run AI features.
Thank you again for your feedback. It really helps us improve both how we communicate and how we design for you.
I just wanted to let you know that this was a known issue which has now been resolved by our dev team. Everything should be working correctly on your side now. Could you please give it a try and let us know if youāre still experiencing any problems?
We really appreciate your understanding, and weāre very sorry for any inconvenience this caused.
Thank you for your response. We were able to resume our operations successfully, and everything is now working as expected.
Iād also like to mention that this change had a strong impact on our scenarios, as we were using many Make AI Modules (as @Petr pointed out). Even so, I personally find the change reasonable ā the timing was just difficult for us. Special thanks to Pavlina M from Customer Care and everyone involved in granting us extra credits, which really helped us continue operations while the issue was resolved.
We truly value this kind of support, and we will gladly continue using Make, knowing that as developers we understand unexpected issues can happen ā and in this case the response was as quick and effective as needed.
Hello again @Robinson_Munoz_Hoyos, thank you so much for your kind words and for sharing your experience with the community. Weāre really happy to hear that your scenarios are able to run as expected, thanks to the help of our Customer Care team.
We truly appreciate your trust, and weāll make sure to pass your thanks on to the team.
Hey, why canāt it be as simple as Zapier, where if a trigger doesnāt receive any data, it doesnāt consume credit? This system of deducting credits even if no action occurs is quite strange, to say the least.
It really helps us improve both how we communicate and how we design 
for you.