Hey everyone! Thanks again for raising these questions about AI credit limits. Good news: we have a few updates for you!
Starting December 9 (and rolling out over the next week), we’re adding more transparency around how AI credits are used across Figma products. You’ll now be able to track your AI credit usage directly in Figma. We’re also expanding the ways you can add more credits next year.
Here’s a summary:
Starting this week, you’ll be able to track your AI credit usage in Figma.
Next year, we’re introducing new ways to buy AI credits:
Starting March 11, 2026, you'll be able to add more credits through an ongoing subscription.
By Q2 2026, you’ll have an option to pay-as-you-go for more credits.
Starting on March 18, 2026, we will enforce all seat credit limits.
For example, toview your seat credits. You can now view how many credits you’ve used, how many credits you have remaining, and when they’ll reset directly in your file menu:
We’ve also introduced a new AI credits page for admins on paid plans to track AI usage across their plan. Read more about managing AI credits as an admin in ourHelp Center.
Question about the Figma Make Credits and Prompt limit. Looking at the data in the link you provided, it looks like users on the Enterprise Plan are allowed 4,250 credits per month. Is that per user or per Enterprise? For example, if our company has an Enterprise account, does each user with a full license get 4,250 credits per month or does the whole company only get 4,250 credits per month?
Full seat AI credits:
Professional plan
Organization plan
Enterprise plan
3,000 credits per month
About 50-70 Make prompts based on current estimates
3,500 credits per month
About 60-80 Make prompts based on current estimates
4,250 credits per month
About 80-100 Make prompts based on current estimates
My project is locked up as well. I am trying to understand if upgrading is an option to unlock. Going from Professional plan to Organizational allows for more credits, but will that unlock my project? Currently I cannot interact with the build in anyway. Sitting on weeks of work that is unavailable is somewhat nerve racking. Upgrading to Enterprise plan indicates that you have to contact sales. But I cannot get a response from web submittal. I am good with the upgrade, but first I need to know that it will unlock my project. My second question for the short term is whether the Enterprise plan is even accessible to upgrade to it?
Hey djonson1223, thanks for reaching out in the Forum! I can see that you’re already in touch with our support team, and based on the latest update on your ticket, your credits have now been reset. In cases like this, the lock could be related to running out of credits, and once credits are restored or the plan is adjusted, access should be unlocked but I’d recommend to check directly with our support team to confirm it. Since this involves account-specific details (credits, plan, and seat configuration), it’s best to continue the conversation directly via your existing support ticket. (The team there can confirm the current status of your project and guide you on any next steps, including upgrade options if needed.) Thank you!
i absolutely love this app. and am building a FULL ERP solution for my existing publishing company. Its amazing. Im a non tech builder.. so im using GPT as well.. and its guiding me and prompting Fig, the agent inside make. We have a good team going . I havent bumped into a credit problem yet. but im scared im going to.
i wish i could see my balance. i also wish there were a bunch of other tools for myself, so i can build context over the long term with my agent, across sessions. Every session i start, i need to update the agent again as to exaclty where we are.. and thats taking tokens im sure. We should get space with you for the fee, to have memories, special instructions , files etc.. that all operate within your rails.
This 3,500 credit / 60-80 prompt limit is borderline useless to a heavy user on an Enterprise Team that designs complex software. I burn through that in 1-2 days.
I have a feature with multiple views and roles I’m building, and already well over 400 prompts, and it’s finally getting close to cleanly simulating the potential MVP. And that is just focusing on content displayed and interactivity/workflow. Final visual design will still need to be done. but will pull those into Design files.
A
Anonymous
Mar 11, 2026 11:50 PM UTC
3000 credits is about 45 mins usage, this has made the platform, which has JUST begun to be useful with the introduction of the more advanced models completely unsustainable from a usability perspective. I will burn 180 credits asking it do do a simple task, the requirement to use a advanced model is basically mandatory as the lesser models just wont get the outcome, The AI agent will forget the context within a prompt or two and will activity LIE to you on the lesser models. I believe the platform has become extremely usable now, but the credit system needs to be reviewed urgently as this is a deal breaker for everyone trying to use it as a actual tool-set. There are alternatives, i suggest everyone start voting with their feet.
A
Anonymous
Mar 12, 2026 1:00 AM UTC
I believe the most fair way to handle this, Figma Make team, is to allow the user in the admin settings to enter their own API keys for their own models, that way the burden of AI model cost is not on you and you will get a recurrent income form supplying the platform, the burden of consumption will then fall upon the consumer as they consume.
As others have said, 3,500 credits won’t last past an hour. Looking at my prompts, there is one prompt that was two small paragraphs long and it used 1,784 credits for 5 minutes of work. For $660/year for 12 hours of use is incredibly expensive compared to the competition. AI is supposed to be much cheaper not on par with a UX engineer. I will unfortunately have to stop using Figma Make when the credits are enforced starting tomorrow on 03/18/2026. For Claude Code $100/month plan, I have it in use at least 5 hours a day and I don’t run out and it creates UI code just as good if not better. If someone from management is listening, you are about to lose a lot of AI customers with this extremely limited usage model.