Figma Make Ai Credit Limits Not Feasible

This post is spot on!  Considering the fact that Anthropic and OpenAI are giving you 10x more credits now for the price than they use to how is 3k credit enough with Figma?  I have another issue, don’t we get 3k credit with a professional seat?  I added 1k more, total of 4k but was shut off today when I used 1k?  Not sure Figma know how to do math.

I recently shared feedback about the Figma Make credit system, but my previous post appears to have been removed without explanation. If it was taken down due to tone or guidelines, I understand moderation is necessary—but I would appreciate more transparency. This is intended as structured, constructive feedback based on real usage.

1. Real Usage Pattern (Heavy User Perspective)

In my case:

• ~5,000+ prompts executed

• Monthly usage previously reached ~100,000–150,000 credits

This was not due to misuse, but because:

AI workflows are inherently iterative — not one-shot.

A typical process looks like:

• Prompt → result → refine → regenerate → repeat

• Often 10–30 iterations are required for a usable outcome

This is normal behavior when working with generative systems.

2. Core Issue: Iteration Is Being Penalized

The current credit model effectively charges per attempt.

However:

“One usable result” often requires multiple attempts.

This creates a mismatch:

• The product encourages rapid iteration

• The pricing discourages iteration

As a result:

The core workflow of exploration → refinement is constrained.

3. Lack of Cost Predictability

According to the documentation:

• Credit usage depends on complexity, model, and internal agent actions

• Cost cannot be estimated in advance

This leads to:

• No way to predict cost before execution

• No ability to plan or optimize usage

In practice:

Cost becomes reactive rather than controllable.

4. Model Selection Without Meaningful Cost Control

Figma Make allows users to select different AI models.

However:

• Credit usage does not appear to be clearly differentiated

• Total cost is still influenced by hidden factors (complexity, agent behavior)

This creates a disconnect:

Model selection is exposed, but cost control is not.

If users are given model choices, it would be more meaningful if:

• Cost differences were transparent

• Users could intentionally trade off quality vs cost

5. Impact on Workflow (Beyond Pricing)

The issue is not just cost.

It directly affects how people work:

• Experimentation becomes risky

• Trying new ideas feels costly

• Users become overly conservative

This leads to:

Reduced exploration and slower product iteration.

6. Strategic Concern

From a product perspective:

• Light users may not notice this issue

• Heavy users (who rely on iteration) are most affected

This creates a risk:

The users who extract the most value may be the ones most constrained.

7. Suggestions

Some potential improvements:

• Provide clearer cost estimation before execution

• Introduce a lower-cost “exploration mode”

• Align model selection with transparent credit differences

• Consider partial cost reduction for iterative refinement within the same context

8. Closing

This is not a complaint about pricing alone.

It is about alignment between:

• How generative AI actually works (iterative, exploratory)

• And how the credit system is currently structured

Additionally, I would appreciate clarification on why previous feedback posts may be removed, so future discussions can remain within guidelines while still addressing real issues.

I’m interested in hearing how other users—especially those working iteratively—are adapting to this system.

I have just bought 2.5K credits. Figma makes so many mistakes and does not correct them without lots of prompts which I am being charged for. This product is not ready to go to market at this price???

Does the point tool in Figma Make need AI credits to be used? I am not generating any prompts. I am not using the AI. It is currently disabled for me because I ran out of credits. Why do I need credits available to use it, if I am not asking the AI for anything?

I just wanted to update text without needing to look through all the code. I was able to use it without triggering a “credit usage” before the updates. I hope this is an error and not intentional, because that would be ridiculous when so many other website building sites let you change the text by just clicking it.

Hi,
We’ve been attempting to scale our UI design process via Figma, but the current AI credit constraints are a significant bottleneck. Despite using precise prompts, the credit allotment is exhausted during the initial scaffolding phase, making it impossible to complete full screen designs or iterations. At this stage, Figma Make isn't viable for our workflow; shifting toward an MCP and AI-driven approach within GitHub appears to be a more efficient alternative.

Albert

Es un abuso he comprado 1000 créditos adicionales y solo he podido hacer tres consultas cuando se agotaron, esto es un robo. Cualquier otro sistema de IA gestiona muchísimo mejor los créditos, creo que a sido un autentico error que deben de corregir de inmediato, no me opongo a comprar créditos si es necesario, pero si tengo que gastar miles de $ en 4 sesiones de IA no le sale rentable a nadie. Hasta ahora he estado super contento con Figma, pero lamentándolo mucho si no cambian la política de gestión de Créditos, tendré que cerrar la cuenta e ir a otra plataforma para mis proyectos. Encima no hay forma de contactar con ellos, no me parece profesional. Yo estaba recomendando a todo el mundo Figma,  pero ahora dejaré de recomendarlo.

My comment on a previous post yesterdat now seems to have been removed and the topic is now closed.

We are suffering a similar situation where we are seriously evaluating the feasibility of using Figma make.  I am a recent user with very little design experience but found make a game changer when making UX prototypes to then collab with our designer.  This increased our productivity at least tenfold.

With the laughable limits now being enforced combined with often fighting the AI we can burn through 3000+ credits in less than an hour trying to get it to fix issues it made and hallucinating that the issue is now fixed.

Currently looking at Claude Artifacts and Google Stitch.

That is all!!!

The current AI credit model isn’t viable for us from a business perspective.

We’ve been actively using Figma AI during the no-cap phase across exploration, iteration, and rapid prototyping. It worked well because it removed friction — teams could experiment freely and integrate AI into daily workflows.

With the new credit structure, usage becomes unpredictable and hard to budget. That uncertainty alone makes it difficult to adopt at scale. We can’t realistically plan or justify spend when costs fluctuate based on experimentation.

For AI to stick in a product/design workflow, it needs to feel like infrastructure — not a metered tool you hesitate to use.

A fixed, predictable pricing model (e.g. “all-you-can-use” with a reasonable daily cap) would make far more sense. That’s how other tools are enabling real adoption — by encouraging usage, not constraining it.

Right now, the model discourages exactly the kind of frequent, exploratory use that makes AI valuable.

The new AI credit limits were introduced yesterday, and by today I have already run out of Figma Make credits. After about four hours of use (spread over two days), it mostly spent time fixing its own bugs, so my impression is that it is not yet usable in its current state.

I understand the need to balance costs and monetization, but the current limits significantly impact usability. I hope this can be adjusted so that Figma Make remains a practical and effective tool.

It’s a shame because despite the constant bugs, we’ve found it massively sped up our iterations. However, the pricing model isn’t sustainable and this is forcing us to consider other market options. We wouldn’t have done this if we could continue working the way we were. Figma needs to thoroughly investigate alternatives and ensure they’re ahead of the curve.