article

Figma AI Credits, Explained: What You Get and What Runs Out

TL;DR Figma AI credits are a per-seat monthly allowance (3,000 on Professional, 500 on Starter) that resets each month and doesn't roll over. Cheap actions like background removal cost a few credits; Figma Make and image generation can burn far more. When you run out, you wait for the reset or buy more at $0.03/credit. Run your own usage through the calculator below before you worry.

If you’re seeing a “You’re low on AI credits” banner in Figma, here’s the short version: every Figma seat gets a monthly AI credit allowance, credits reset each month and don’t roll over, and different AI features spend wildly different amounts. Here’s exactly how many you get:

PlanFull seat credits / month
Starter (free)500 (plus a 150/day cap)
Professional3,000
Organization3,500
Enterprise4,250

That’s the number most people are hunting for. The rest of this guide covers what a credit actually is, what burns them fastest, what happens when you hit zero, and how to buy more — plus a calculator to estimate your own month.

Figma’s own docs have all of this, but it’s scattered across five separate help articles (AI credits, billing, pricing, AI models, product updates). This pulls it into one reference.

Take Mina, a freelance designer, who saw that banner halfway through a pitch deck. She had three questions:

This guide answers all three.

What an AI credit is (30 seconds)

A credit is the unit Figma charges against when you run an AI action — Figma Make, AI image editing, agentic features, and so on. Think of it less like money and more like a monthly book of tickets: you get a stack, each AI action tears off a few, and the stack refills at the start of your billing month.

A few rules that trip people up (per Figma’s help docs):

That’s the whole concept. The part people actually search for is the next two sections: how many you get, and what burns them.

The seat-type catch that surprises people

The table above is for a Full seat — the standard design seat. But your allowance depends on your seat type as much as your plan, and this is where the forums fill up with confusion.

Dev, Collab, and View seats get 500 credits/month on every plan — even Enterprise. So if you’re on a Dev seat wondering why you have a fraction of your designer teammate’s credits, that’s why: upgrading the org’s plan does nothing for a non-Full seat. Only Full seats scale with the plan (500 → 3,000 → 3,500 → 4,250).

Starter users and View seats also hit a daily cap of 150 credits on top of the monthly number, so a heavy afternoon can stall you until tomorrow even if you have monthly credits left.

What actually eats your credits

Not every AI action costs the same, and some are free. Here’s Figma’s published consumption for the common features (rates as of July 2, 2026 — Figma notes these can change as models are optimized):

FeatureCredits per use
AI search, Rename layers, FigJam summarize/clusterFree
Remove background1–5 per image
Vectorize image2–5 per image
Boost image resolution5–10 per image
Erase / isolate / expand object5–10 per image
Make / edit image2–16 (depends on model)
Add interactions20 per use
Figma Make, agentic featuresVariable

Two things worth pulling out of that table:

Image model choice can be an 8× swing. “Make image” runs anywhere from 2 credits (ChatGPT Images 2.0) to 16 (Nano Banana 2), for the same click. If you’re generating a lot of drafts, picking a cheaper model for rough passes and saving the expensive one for finals is the single easiest way to stretch your allowance.

Figma Make and agentic features are the real budget sink. They’re billed variably — a single involved run can consume many times what a background removal does, and there’s no fixed per-use number to plan around. That’s usually the culprit when someone burns through credits faster than they expected.

What 3,000 credits looks like in practice

The official rates tell you what each action costs, but not whether you’ll run out. So here are three realistic monthly patterns, worked out from Figma’s per-action numbers above. (These are our estimates built from the official rates, not Figma figures — your real usage will vary, and the calculator below lets you plug in your own.)

A freelance designer doing light image cleanup. 15 background removals + 10 resolution boosts a month works out to 15 × 3 + 10 × 8 = 125 credits, call it 125–150 with a few extras. That’s comfortable even on the free Starter plan (500), and barely a dent in Professional’s 3,000. At this level you’re unlikely to hit the monthly limit at all.

A marketing team leaning on image generation. Two people generating 40 images a month each on a mid-tier model (×8), plus some edits: 2 × 40 × 8 = 640, roughly 1,300 credits once edits and cleanup are added. Safe on Professional’s 3,000 per seat — but notice how model choice swings it. Run those same images on the top model (×16) and the base doubles to 2 × 40 × 16 = 1,280, pushing the total toward 2,600 and close to the ceiling.

A product team using Figma Make daily. This is where it gets slippery. Make is billed variably, so there’s no clean multiplication to do — but daily agentic generation is the usage pattern most likely to exceed the 3,000 included credits and need a top-up. If your workflow is Make-heavy, assume the included allowance won’t cover it and plan for a subscription.

The pattern across all three: fixed image actions are predictable and cheap; Make and agentic runs are the ones that put you over. Which is exactly why the calculator further down separates them out.

What happens when you run out

Nothing dramatic — but the AI actions stop. Once you hit your monthly limit, credit-consuming features are blocked until your next monthly reset date. The free actions (AI search, rename layers, FigJam summarize) keep working; only the credit-metered ones pause.

Because credits don’t roll over, there’s no “saving up” for a big month — each month starts fresh at your plan’s allowance. If you’re on Starter or a View seat and hit the daily 150 cap, you’re back in business the next day without waiting for the monthly reset.

You’ll sometimes see credits referred to as your “balance” in the billing dashboard — same thing, just the remaining count for the current cycle.

Need more? Subscription vs pay-as-you-go

Since March 11, 2026, Figma offers two ways to buy credits beyond what your seats include (announced on the Figma blog):

The pay-as-you-go rate is $0.03 per credit (as of July 2026). That sounds trivial — a background removal is pennies — but remember Figma Make and agentic runs consume credits in bulk. A few heavy sessions can add up faster than “three cents each” suggests, which is exactly why the spending limit exists. Set one.

Rates and allowances change; the numbers above are Figma’s official figures as of July 2026. For your exact current balance and pricing, check your Figma billing dashboard.

So how much will you actually use?

A Professional seat gives you 3,000 credits, which sounds like plenty. But if you generate 20 images a day, edit half of them, and reach for a premium model, you can be at the ceiling well before month’s end — as the marketing-team scenario above showed, one model choice can double your burn.

The scenarios above use our assumptions. The calculator lets you replace them with your own workflow. Plug in a typical month — it uses Figma’s published per-action rates and shows where you’d land against your plan’s limit, and if you’d go over, roughly what pay-as-you-go would cost.

How many credits will you actually use?

Enter what you run in a typical month. Credit costs are Figma’s official rates (as of July 2026). Nothing leaves your browser.

Remove background1–5 / image
/mo
Boost image resolution5–10 / image
/mo
Erase / isolate / expand object5–10 / image
/mo
Vectorize image2–5 / image
/mo
Add interactions (prototype)20 / use
/mo
Make / edit imagemodel-dependent
/mo
0 credits / month0% of your 3,000 limit

Add a few numbers above to see where you land.

Figma Make and other agentic features are billed variably per run and aren’t in this estimate. Check your Figma billing dashboard for exact, current numbers.

If it says you’re over, the fastest fix usually isn’t buying more — it’s dropping to a cheaper image model for rough passes and reserving the expensive one for finals. Much of the time, “AI credit anxiety” comes down to running a premium model on throwaway drafts.

If you’re comfortably under, the banner that started all this was just Figma being cautious — you have room to spare.

Quick answers

Not sure which plan or workflow fits your team? Get in touch — happy to compare notes.