Muse Image Generator

Make quick drafts with Muse Image, Meta’s image model, for 2 credits each: stylized looks, short bold text and up to 10 references, with credits shown first.

  • Made by Meta
  • 2 credits an image
  • One size, about 2.5 MP
  • Four more models in the same menu

Muse Image

Muse ImageAI enhance
0 / 4000

Made with Muse Image

Real AI-generated examples, each made with Muse Image in this generator. Click one to fill in its prompt, text and settings, then change what you like.

  • Yogurt tubes box for DAIRY DOODLES Strawberry Yogurt Tubes, an AI-generated example made with Muse Image
  • Energy drink can for HYPERDRIVE Cosmic Berry Energy, an AI-generated example made with Muse Image
  • Snack bag for PIXEL PUFFS Cheddar Cheese Puffs, an AI-generated example made with Muse Image
  • Gingerbread kit box for TWINKLE TOWN Gingerbread House Kit, an AI-generated example made with Muse Image
  • Sock header card for SOCK RIOT 3-Pack Crew Socks, an AI-generated example made with Muse Image
  • Juice bottle label for PULP & PRESS Green Juice, an AI-generated example made with Muse Image
  • Toy box for STACKTOWN City Builder Set, 420 Pieces, an AI-generated example made with Muse Image
  • Bonbon box for TOPSY TURVY CONFECTIONS Salted Caramel Bonbons, an AI-generated example made with Muse Image

Muse Image is made by Meta. Pack AI Studio runs it through fal.ai and is not affiliated with or endorsed by Meta. Image generation is part of the paid plans, and credits for a failed run come back. Keep printed text short and check every word before you print.

credits an image, the lowest in the menu
2
one fixed size; only the shape changes
2.5 MP
reference images in one run
10
aspect ratios, from 21:9 to 9:21
11

How to use Muse Image here

The generator opens on Muse Image. Three steps, and the button shows the credits before anything is charged.

What Muse Image does well

Cheap, quick and bold, which is what a first round of ideas needs.

Energy drink can for HYPERDRIVE Cosmic Berry Energy, an AI-generated example made with Muse Image
AI-generated example · Muse Image

Drafts for 2 credits

Four Muse Image drafts cost 8 credits, no more than a single 1K image on any other model here. Try ten directions, then spend more on the one you keep.

Gingerbread kit box for TWINKLE TOWN Gingerbread House Kit, an AI-generated example made with Muse Image
AI-generated example · Muse Image

Named styles, done boldly

Papercut, claymation, Y2K chrome, gouache and 8-bit pixel art come through clearly when you name them in the prompt.

Toy box for STACKTOWN City Builder Set, 420 Pieces, an AI-generated example made with Muse Image
AI-generated example · Muse Image

Short text that holds

A brand name, a product name and a short line such as "Ages 6+" are the text it handles best. Long lines and small print are where it slips.

Sock header card for SOCK RIOT 3-Pack Crew Socks, an AI-generated example made with Muse Image
AI-generated example · Muse Image

Up to 10 references

Add a logo, a product photo or a style you like. Meta built Muse Image to combine elements from several reference images in one picture.

You type

Packaging for kids yogurt tubes in claymation style: chunky clay cows and strawberries, soft pastel pink and green, playful.

AI enhance writes

Flat print-ready packaging artwork for a kids yogurt tube box, claymation style with chunky hand-sculpted clay cows and strawberries, soft pastel pink and mint green palette, playful and cheerful mood. Front panel layout: a soft pastel pink background with a mint green rolling hill band along the bottom. Centered at the top, the brand name "DAIRY DOODLES" in bold, rounded, clay-textured letters. Below it, the product name "Strawberry Yogurt Tubes" in a smaller friendly playful font. Chunky clay cows with squishy fingerprint textures and clay strawberries with tiny seeds sit beside the text and along the hill, evenly spaced and not overlapping the lettering. In the lower corner, the net quantity "8 x 2 OZ (57 g)" in a clean small label-style font. Soft even studio lighting, straight-on orthographic view, no perspective, no shadows falling on the text, matte clay finish, clean edges for die-cut printing.

AI enhance

A short request becomes a full brief: subject, setting, style, palette and where each line sits. Quoted words and the text fields stay as you typed them.

  • GPT Image 2.5 FlarePrecise text, fasterfrom 10 credits
  • GPT Image 2.5 SunburstPrecise text, finer detailfrom 10 credits
  • Nano Banana 2.1Many references, extra widefrom 8 credits
  • Seedream 5.0 ProPhoto real, high resolutionfrom 11 credits
  • Muse ImageLow cost draftsfrom 2 credits

Switch when it matters

When a draft is right, run it again on GPT Image 2.5 for exact text, Seedream 5.0 Pro for photo real detail or Nano Banana 2.1 for edits.

What Muse Image is, as of October 2026

Muse Image is the first image model from Meta Superintelligence Labs. Meta launched it on July 7, 2026 in the Meta AI app and on meta.ai, with Instagram and WhatsApp features built on it, and CNBC reported that it was code-named Mango inside the company.

Meta describes it as working like an agent: according to the AI at Meta blog, Muse Image calls search and coding tools, refines its own output, writes code to draw accurate charts and QR codes, and composes elements from many reference images.

We run Muse Image through fal.ai, where it is priced at a cent an image, which is why it costs only 2 credits here. Facts on this page were checked against Meta, CNBC and fal pages in October 2026.

Bonbon box for TOPSY TURVY CONFECTIONS Salted Caramel Bonbons, an AI-generated example made with Muse Image
AI-generated example · Muse Image

Muse Image specs and credits on Pack AI Studio

Credits: 2 an image, 1 to 4 images a run. Size: one fixed size of about 2.5 megapixels; you choose the shape, not the resolution, so the resolution menu shows a single option. Shapes: all 11 aspect ratios from 21:9 to 9:21. References: up to 10. Time: usually 10 to 30 seconds, the quickest range in our menu.

In our own test on October 9, 2026, Muse Image made a packaging design in 18 seconds and printed "MOSS & CO" and "NET WT 4 OZ (113 g)" correctly. You get a PNG per image. Credits come with the plans on pricing, and a run that fails returns its credits.

Snack bag for PIXEL PUFFS Cheddar Cheese Puffs, an AI-generated example made with Muse Image
AI-generated example · Muse Image

Where Muse Image falls short

Small text. Reviewers found that fine type can come out garbled: DataCamp wrote on July 8, 2026 that small text on signs, interface mockups and product labels can come out garbled. Big words, such as a brand name and a product name, hold up far better.

So keep the text to print to a few short lines with Muse Image, and leave long copy, ingredient lists and fine print to a model that handles them, or to the design file. The fixed size is the other limit: at about 2.5 megapixels it is fine for screens and drafts, while a large print may need a 2K file from another model.

Juice bottle label for PULP & PRESS Green Juice, an AI-generated example made with Muse Image
AI-generated example · Muse Image

Muse Image compared with the other models here

GPT Image 2.5 Flare and Sunburst (OpenAI) cost 10 to 15 credits and are the pick for exact, small text. Seedream 5.0 Pro (ByteDance) costs 11 to 22 credits and is strong on photo real materials and dense layouts. Nano Banana 2.1 (Google) costs 8 or 11 credits and is built for edits from photos; it has no 9:21.

Muse Image costs 2 credits and is our pick for exploring: styles, palettes, layouts and names before any of them is final. See the other models on Seedream 5.0, GPT Image 2 and Nano Banana 2.

A draft first routine for packaging

Run four Muse Image drafts of the same brief in different styles: 8 credits. Keep the best one, click Use as reference, then run it on a finishing model with the full text in the fields. The draft keeps the look; the second model gets the words right.

Type the net quantity the way it must print. On US food packages it goes in the bottom 30% of the front panel in bold type (21 CFR 101.7), in both US and metric units (15 U.S.C. 1453), such as "NET WT 7 OZ (198 g)".

Build exact parts with free tools rather than the model: barcodes with the UPC barcode generator, Nutrition Facts with the nutrition label generator and the box with the dieline generator. To design on the dieline from the start, use AI packaging design, where Muse Image is in the menu too.

Who uses Muse Image

Anyone who wants many ideas before paying for one finished image.

Brands choosing a style

Put claymation, papercut and pixel art side by side for a few credits.

Designers before a pitch

Show a client a wall of directions, then finish the winner on another model.

Kids and toy products

Playful boxes and cards with a short name and an age line.

Seasonal and gift lines

Quick looks for limited editions where the style matters more than the small print.

Muse Image questions

What people ask about the model and about running it here.

Generate with Muse Image

What is Muse Image?

Muse Image is Meta’s image generation model, the first from Meta Superintelligence Labs, launched on July 7, 2026. It makes and edits images and can combine several reference images.

Is Muse Image free?

Meta offers it free in the Meta AI app with limits. Here each Muse Image costs 2 credits, and credits come with the plans on the pricing page.

Is Muse Image good at text?

With short, large words, mostly yes. Testers report that small text can come out garbled, so keep printed lines short and check every word.

What resolution does Muse Image make?

One fixed size of about 2.5 megapixels. You pick the aspect ratio, from 21:9 to 9:21, and the size follows.

Can Muse Image use my own images?

Yes. Add up to 10 reference images, such as a logo, a product photo or a style you like.

When should I use another model instead?

Use GPT Image 2.5 for many lines of exact text, Seedream 5.0 Pro for photo real detail, and Nano Banana 2.1 to edit a photo of your product.

Is Pack AI Studio part of Meta?

No. Muse Image is made by Meta; we run it through fal.ai with four other models, and we are not affiliated with or endorsed by Meta.

What content is not allowed?

No nudity, sexual content or gore, in prompts, uploads or results. Such requests are refused before any credits are charged.

Try Muse Image now

Describe a style and a subject, type a few short words and make four drafts for 8 credits. The generator starts on Muse Image and shows the credits first.