Figma Make Ai Credit Limits Not Feasible

​@mbanks very easy to eject your figma make projects locally and use claude code which is an order of magnitude more powerful - or just use figma for the first few design screens  - get that dialed in and switch over.  I find claude code to be equal and in some cases superior when it comes to design and much faster when it comes to resolving bugs. I put some instructions a few rows up but all depends on what you’re trying to do. in order to share designs i did need to figure out a solid hosting solution so setup a git, vercel, and supabase account (all free for light usage) - claude code walked me through it all.  Very easy. 30-60 minutes and you’re off to the races.  Also using Codex a bit now too connected to the same project.

We would like to provide feedback regarding the current AI credit consumption policy.

In its current form, the credit system does not feel fair or aligned with real user value. While AI features can be helpful in producing design ideas or accelerating workflows, the current implementation consumes credits at a relatively high rate—even in cases where the generated output is not fully usable or requires multiple iterations to reach an acceptable result.

This creates a situation where users are effectively charged not only for successful or meaningful outputs, but also for incomplete or suboptimal generations that still require additional prompts and refinement. As a result, the cost of experimentation and iteration becomes disproportionately high, especially for designers who rely on trial-and-error workflows.

We believe this approach may negatively impact long-term user experience and satisfaction. Many long-time users who have been part of the Figma ecosystem for years may reconsider their continued usage if AI features remain tied to such restrictive consumption mechanics.

AI should enhance creativity and productivity—not introduce friction or hesitation in the creative process. When users begin to limit their exploration due to credit concerns, the core value of these tools becomes diminished.

We strongly encourage reconsideration of the current credit consumption model, either by:
- reducing credit cost per iteration,
- improving the accuracy and usefulness of generated outputs,
- or introducing more forgiving evaluation metrics that distinguish between low-quality and high-value generations.

Otherwise, there is a real risk that users may gradually migrate to alternative platforms that offer more flexible or transparent AI usage models.

We appreciate the effort behind integrating AI into Figma, but believe that a more balanced and user-centric credit policy would significantly improve adoption, trust, and long-term engagement.

[Moderator note: This post was merged from Credit consumption into Figma Make Ai Credit Limits Not Feasible on 12-Apr-2026.]

I wrote the exact same feedback, I haven’t touched my figma page in over a week now, because I have to wait until April 29th to continue my project, because I ran out of my “20 000” credits! Can you beleive this, I ran out of 20 000 credits and now I have to wait 3 weeks to continue my project. You think my energy and motivation level will be the same in 3 weeks? Not happy right now. Don’t get me wrong, I love Figma and the community, but I feel they are not listening to us. :(

[Moderator note: This post was merged from Credit consumption into Figma Make Ai Credit Limits Not Feasible on 12-Apr-2026.]

Hi ​@arsalan moqadam and ​@Ron Raymond, thank you both for taking the time to share this. I hear you on your core points you're making about: the cost of iteration and the concern about experimentation.

I've merged your posts into a thread where we're consolidating feedback on AI credit consumption and keeping the loop open with our product teams. Please take a look at the pinned reply from ​@ksn for more context on where things stand.

For reference, this resource is kept up to date as the credit system evolves: How AI credits work.

1000/3500 credits used in less than 2 hours on the 1st of a new month. Claude stuck in a loop of fixing an unknown error where my app won't load. Each "fix" was easily 100 credits. AI was clearly not made to handle a sandbox environment. Note that Claude “auto fixes” errors. I was out to lunch when it looped 10x and wasted 1000 credits before pausing.

Bye Figma Make. You win, I’m done.

As a designer managing multiple complex projects, I find the current AI credit system in Figma make to be a major bottleneck. Currently, 10–12 prompts in a single file can exhaust the limit in just one day, leaving the tool unusable for the remaining 29 days of the month. This isn't just frustrating; it’s a productivity killer.

The Core Issues:

• Iteration vs. Luxury: Most AI usage isn't for "one-click magic"—it’s for fixing incomplete outputs and refining layouts. Penalizing these necessary iterations with credits makes the tool feel like a liability rather than an accelerator.

• The "29-Day Deadzone": If I hit my limit on day one, I can’t use Figma make for the rest of the month. This makes it impossible to rely on the tool for professional, time-sensitive projects that require daily updates.

• Predictability is Zero: In a pixel-perfect workflow, I can't predict how many prompts I’ll need to get a prototype right. Running out of credits mid-way before a stakeholder review forces me to manually rebuild everything, defeating the purpose of AI automation.

• Stakeholder Friction: Figma make is my bridge to non-designers. When credits run out, I lose the ability to share functional, clickable links, leading to massive misalignment with my team. 

My Recommendation:

Instead of a rigid monthly cap that punishes power users, I suggest a more sustainable model:

Daily/Periodic Refresh: Implement a limit that refreshes every 24 hours (similar to Claude’s model). This ensures that even if we hit a wall today, we can continue our work tomorrow.

Unlimited for Paid Tiers: Professional/Organization plans should come with the expectation of scalability. Consider unlimited usage for paid users, with limits only to prevent bot abuse.

Local Refinement: Allow users to refine previously generated AI elements without consuming new credits for every minor tweak.

Conclusion: Design is an iterative process. A credit-based system that stifles exploration and penalizes professional usage will only drive designers back to traditional, manual workflows. We need a system that supports the way we actually work.

#Figma #Figmake #AI #UXDesign #ProductFeedback

I 100% agree! As it stands, this feature has become more of a problem than a solution.

This is obivously their new business model to generate profit, which is okay, but there is no clarity and accountability. 

When I asked support why does adding a single word cost 30 credits and adding 5+ rows with paragraphs of information cost 46 credits they simply give a politician-style non-answer: 

“Thank you for your reply.
 
AI credits are consumed based on the work Figma Make performs in response to a prompt. In Figma Make, an “output” includes any work the model completes such as generating or updating code or making changes to your file. If any such changes are made, including any follow up prompt to refine results, credits will be consumed based on the work performed. (For example, even if a preview doesn’t load, credits will still be used if work was performed.) Any follow up prompts to refine results will also consume credits based on the work performed.
 
If a prompt fully fails and no work is performed or output was generated, credits are not deducted.
 
Hope this helps, and please let us know if you have any further questions.”

It just doesn’t make sense…. I added 2 pics and replaced 2 pics to the bottom of a row on roughly 12 pages and the cost was anywhere from 56 credits to 104. The pics are all the same size so it makes no logical sense other than, “we want more of your money.”

I’m glad that there are similar alternatives that are much, much cheaper. 

I agree the current credit limit is unworkable for most projects, fixing basic design issues eats into credits, it’s knowledge of frontend development seems weak in some areas, it could not work out how to get the header to appear in front of the scrolling page content, it struggled with laying out content in a table format even though it had a visual reference to work from, it ignored instructions when I said I wanted to reskin a referenced UI rather than redesign and improve, I barely completed a 1 page design with 1000 credits that was based on an existing design visual which still needs to be reviewed by another team, we will then need further iterations on this before sign-off, and that’s before any other pages have been created. Just glad this is for an internal project rather than for a client - telling teams and clients I can’t complete the project due to a credit limit is not workable.

The other issue is that we might not use Figma Make for months, so perhaps having credits rollover each month instead of resetting would be a better solution, or allowing users in the same team to share credits with another user seems a fairer model and may work for us.

I echo all the comments,  this model is not sustainable as AI becomes a large initiative for many companies. the per user model should be adjusted to a Org AI credit model.  There are lots of credits sitting on the table that are unused.