What Is Figma AI? Features and What It Can't Make

Figma AI is the set of AI tools built into Figma: an agent that generates and edits designs, Figma Make for prompt-to-app prototypes, and image tools for making, editing and cleaning up pictures. As of October 2026 its AI features share one pool of credits, and what you get depends on your plan and seat.

What Figma AI covers

Figma's help center sorts its AI into four groups, and it helps to keep them apart.

• The Figma agent is a chat tool. It generates and edits designs, applies your design system, gives feedback and takes on repetitive tasks. Figma says it can also create custom plugins and shaders and generate animations. It works in Figma Design; in FigJam and Figma Slides it is limited to closed beta participants.

• Figma Make turns a prompt or an existing design into a working prototype or web app, in its own Make file. Apps can be published as websites.

• Code layers (open beta) put working code on the Figma Design canvas, so you can compare interactive prototypes next to your designs.

• Individual tools do single jobs: rewrite or translate text, make or edit an image from a prompt, remove a background, expand an image, isolate or erase an object, vectorize, boost resolution, rename layers, add interactions and search for existing designs and assets.

Figma Weave sits next to these. Weave is the new name for Weavy, which Figma acquired and announced in October 2025. It runs as its own node-based canvas at weave.figma.com and lets you choose a model for each task, and Figma's launch post lists images, video, animation, motion design and VFX among the media it covers. Weave tools for variations, retouching and styles can also run inside Figma Design. Weave has its own credits, separate from Figma AI credits.

Plans, seats and credits

As of October 2026, Figma meters its AI in credits shared across its AI features. Each Full seat gets a monthly amount: 500 on Starter, 3,000 on Professional, 3,500 on Organization and 4,250 on Enterprise. Dev, Collab and View seats get 500 on Starter, Professional and Organization. Starter and View-seat users are also capped at 150 credits a day. Credits expire monthly and do not roll over, and at the limit you cannot run more AI actions until credits are available again.

Plan rules decide which tools you see. On Starter the individual image tools work in Figma Design only. Most other standalone tools need a paid plan, though most are still reachable through the agent, which is on every plan. Figma Make gives Full seats the full feature set, and Starter gets it with limitations.

The rates table on Figma's credits page, dated October 15, 2026, puts numbers on the work. Making or editing an image costs 2 credits with ChatGPT Images 2.0, 6 with Nano Banana 2 Lite, 8 with Nano Banana and 16 with Nano Banana 2. Bigger jobs cost more, and Figma gives them as approximate figures: about 100 credits or more to generate an app from scratch in Figma Make, and about 300 or more for a five-screen checkout flow from the agent.

What Figma says it can't do on its own

These are the limits Figma's own pages state, and they are the ones to know before you plan work around it.

• A Make app is a front end first. Keeping app state across browsers, login screens, saving content and storing secrets or API keys all need a backend added.

• Seats matter. In shared Figma Design files, the AI tools need a Full seat with edit access. Dev, Collab and View seats, or view-only access, can only ask the agent questions.

• Some groups get less. K-12 Education users have no Figma AI features, Figma for Government has Figma Make only, and admins can turn AI off for a whole team.

• Credits cap the volume. A five-screen checkout flow from the agent costs about 300 credits or more, against 500 a month for a Starter seat.

• You check the output. Figma tells users to verify AI results before relying on them and to disclose AI-generated content.

Video comes through Weave, a separate product. Its image-to-video page names Kling, Runway, Luma and Wan, and its actions use Weave credits, not Figma AI credits.

Making the images a Figma file needs

Inside Figma it takes a few clicks. In Figma Design, FigJam, Draw, Slides or Sites, open the Actions menu, choose Make an image, write a prompt, pick a model from the dropdown, add a reference image if you have one and send. Select a layer first and the result fills it; deselect everything and it lands as a new image layer. To change a picture already on the canvas, select it and use Edit with prompt. That is the shortest route when the picture lives and dies inside the file.

A separate generator makes sense when the picture is the job and a set has to keep one face from image to image. The model families overlap with Figma's own list. AIGE puts Nano Banana 2 and Pro, GPT Image 2 and Recraft 4.1 in one picker, and the credit cost shows before you press Generate. A 1K Nano Banana 2 image costs 8 credits, and GPT Image 2 on its low setting costs 1 credit. A saved character keeps the same face from image to image; the AI character generator page explains how, and the AI image generator page has the full price table.

When an image is ready, download it and drop it onto the Figma canvas. Figma places a dropped image as a rectangle layer with the image as its fill. Text inside an image is the place to check twice. AIGE's own page names GPT Image 2 as the best bet for lettering and still says to check before you publish, and the same prompt can give different results from run to run.

Honest note: AIGE is not a design tool and does not replace Figma. It makes the pictures that go into one. It has no free generation: making images needs an account and a paid plan. The Figma details above are as of October 2026 and come from Figma's help center; credit rates and plan rules change, so check your own file before relying on a number.

Quick answers