Skip to content
trending

Benchmarks

Meta Muse Image: Price, Availability & Test Results

Meta's Muse Image ranks 7th in Artificial Analysis at $0.01 per image. We tested it with five prompts and it scored 8.9/10, beating models 13x more expensive.

By 7 min read
Meta Muse Image: Price, Availability & Test Results

👁 119 reads

Muse Image is Meta’s image generator, the first graphics model from the Meta Superintelligence Labs team, available since July 7, 2026 in Meta AI. It ranks seventh in Artificial Analysis and costs $0.01 per image via API, the lowest price in the top ten. We checked where you can use it, whether it works in the EU, and how much it costs. On September 27, 2026, we ran it through five prompts from our image generator ranking. It scored 8.9/10, the highest of all lesser-known models we tested that day.

Summary

Muse Image has 1111 Elo and 7th place in Artificial Analysis, and costs $0.01 per image, half the price of the cheapest competitor in the top ten. For regular users, it works in the Meta AI app and on meta.ai, in Instagram Stories in the US, and in WhatsApp in selected countries. Meta has not announced a date for the European Union, and the European version of Meta AI has not generated images since its 2025 launch. Developers have Meta Model API, fal, and OpenRouter. In our test, Muse Image scored an average of 8.9/10, correctly wrote “PIEKARNIA ŻÓŁĆ“ (Bakery Bile), and generated images in an average of 16.5 seconds. It beat GPT Image 2.5 Sunburst in “max” quality (8.8) at 13 times lower cost. Only Nano Banana Pro from the ranking scored higher (9.0), which costs $0.13 per image. One flaw: it ignored the 16:9 aspect ratio and returned every image in 3:2.

8.9/10average in our 5-prompt test
$0.01price per image
16.5 saverage time per image
1111Elo, 7th place in Artificial Analysis

What is Muse Image

Meta describes Muse Image as an “agentic” model. Before drawing an image, it breaks down the prompt into steps, can access a web search and image search or run code, then refines its own sketches. It works with the Muse Spark language model. It generates and edits images from text and from multiple reference images at once. Previously, Meta AI used models from external companies, including Midjourney and Black Forest Labs.

Don’t confuse Muse Image with the Meta Muse app. That’s a separate product, an AI assistant that launched on September 8, 2026, only in the US. We covered it in our article Meta Muse: new AI assistant.

The launch didn’t go without a hitch. In the first version, you could tag a public Instagram account and use its photos as references without notifying the owner. On July 10, 2026, three days after launch, Meta withdrew this feature. Details in our articles Meta Muse: generator that can use your photos and Meta withdrew the Instagram feature.

Where to use Muse Image

Platform Availability Price
Meta AI app and meta.ai since July 7, 2026, excluding European Union free with limit, more in paid plan
Instagram Stories US only free
WhatsApp selected countries free
Facebook, Messenger announced -
Meta Model API for developers $0.01 per image
fal API, generation and editing $0.01 per image
OpenRouter API, requires 18+ age verification $0.01 per image

Advertisers will get the model in Advantage+ creative tools. Meta states that daily image creation in Meta AI is free, and paid plans offer higher limits. It doesn’t publish specific numbers.

Does Muse Image work in the EU

In the app, probably not. Meta AI launched in the European Union in March 2025, but with limited features: without image generation that Americans had from the start. The Muse Image announcement from July 7, 2026 doesn’t mention a date for Europe, and German site Servola reports that image generation is a category Meta is withholding in the EU due to GDPR, DSA, and the AI Act. We haven’t tested the Meta AI app from an EU account ourselves, so we’re relying on these sources.

API is a different story. Meta Model API documentation doesn’t mention regional restrictions, and through fal and OpenRouter the model is available like any other. On OpenRouter, however, you must first confirm in your account settings that you are 18 or older.

How much does Muse Image cost

Meta charges a flat rate of $0.01 per generated image. The price doesn’t depend on prompt length, “reasoning” strength, or whether the model used search. You only pay for images that were returned: those rejected by safety filters are free. The limit is 150 requests per minute. Here’s how it compares to the Artificial Analysis top ten:

Price of 1000 images in Artificial Analysis (USD, lower = cheaper)
Muse Image10
MAI-Image-2.6-Flash19.5
MAI-Image-2.638.9
MAI-Image-2.548.1
Grok Imagine Image 2.060
Nano Banana 267
GPT Image 1.5 (high)133
GPT Image 2.5 Sunburst (max)210.7
GPT Image 2.5 Flare (max)210.7
GPT Image 2 (high)211

Muse Image is 21 times cheaper than GPT Image 2.5 in “max” quality, while losing 85 Elo points to it. In our test it was the opposite of Elo: Muse Image scored 8.9/10, while Sunburst in “max” got 8.8. In our opinion: at this price, it’s the most interesting model from the top tier for mass production via API.

Our test: Muse Image on five prompts

We used the same script and the same prompts as in our ranking: a market, a sign reading “PIEKARNIA ŻÓŁĆ” (Polish for “Bakery Bile”), a flat illustration with a llama, a portrait with hands, and pierogi from above. We tested it in Polish, a heavily inflected language with diacritics, to evaluate non-English text rendering. One attempt per prompt, no cherry-picking, through OpenRouter’s image API (/api/v1/images). Each image cost exactly $0.01, and the entire test cost $0.05. Time ranged from 9.7 to 31.6 seconds, averaging 16.5 seconds. The model worked longest on the illustration, which fits the model’s description of planning the image before drawing.

Prompt Score Notes
Market 9 regular market instead of postcard with Palace of Culture, side morning light, shallow depth; cards “POMIDORY 6 zł/kg“ (tomatoes $1.56/kg), “OGÓRKI 5 zł/kg“ (cucumbers $1.30/kg), “KOPER 2 zł/pęczek“ (dill $0.52/bunch) correct
Sign “PIEKARNIA ŻÓŁĆ“ 9.5 all letters correct, ornate gold sign, warm light from window, no added signs
Illustration 8.5 navy and orange, small llama, no text; line work more like linocut than flat press illustration
Portrait 9 freckles on face and hands, gray beard, steam over coffee, both hands correct
Pierogi 8.5 top view, pan-fried pierogi, sour cream with dill, linen napkin; onion rings scattered on table
Average 8.9 highest of six lesser-known models, behind Nano Banana Pro (9.0)
Elderly woman in sweater at market stall with tomatoes, cucumbers and dill, with price cards in Polish
Market (9/10): the most documentary-style shot of the six lesser-known models.
Close-up of price cards reading POMIDORY 6/kg, OGÓRKI 5/kg and KOPER 2/pęczek (Polish text) from Muse Image output
Cards zoomed in, all Polish characters in place.

At the market, Muse Image wrote the same prices as GPT Image 2.5 in “high” quality: tomatoes 6, cucumbers 5, dill 2. We don’t know why this convergence occurred. In our opinion: such amounts simply appear frequently in materials that models train on.

Storefront of Krakow bakery with green sign reading PIEKARNIA ŻÓŁĆ (Polish) generated by Muse Image
Sign (9.5/10): one inscription, without added slogans and signs.
Zoomed sign PIEKARNIA ŻÓŁĆ from Muse Image with correct Polish diacritics: dot over Z, stroke over O, crossed L and stroke over C
Sign zoomed in: dot over Ż, stroke over Ó, crossed Ł and stroke over Ć - all Polish diacritics correctly rendered.
Portrait of man with freckles and gray beard holding steaming coffee mug with both hands, generated by Muse Image
Portrait (9/10): freckles are even on the hands.
Illustration of woman working at night at laptop with orange llama on desk, in navy-orange palette, from Muse Image
Illustration (8.5/10): palette is correct, but hatching resembles linocut more than flat press illustration.
Plate of pan-fried pierogi with onion and sour cream with dill on wooden table, from Muse Image
Pierogi (8.5/10): top-down photo as requested, but onion rings scattered across entire table look staged.

One technical flaw: Muse Image ignored the aspect_ratio parameter. Instead of 16:9, it returned every image in 1920×1280 resolution, a 3:2 aspect ratio. In OpenRouter, the model doesn’t list aspect ratio or resolution among supported parameters, so you need to crop to 16:9 format yourself. On comparison charts in our lesser-known image generators test, we center-cropped it like the other models. We didn’t count this against the score because aspect ratio wasn’t in the prompt content.

Who is Muse Image for

  • Developers with large scale: $0.01 per image with search included is the lowest rate in the Artificial Analysis top ten, and in our test the quality matched models 13 times more expensive.
  • Graphics with non-English text: sign and price cards without errors, without adding slogans nobody asked for.
  • Product catalogs and editing: Meta advertises the model as a tool for changing exactly what you ask for without touching the rest of the image. We didn’t test this.
  • Regular user in the EU: for now there’s no way to use it in Meta AI. Free alternatives are in our free image generators roundup.

Frequently asked questions

Is Muse Image free?

In the Meta AI app yes, with a limit, but Meta AI in the European Union doesn’t generate images. In the API, Muse Image costs $0.01 per image.

Does Muse Image work in the EU?

According to available sources, not in the Meta AI app: Meta hasn’t announced a date for the EU. Through API (Meta Model API, fal, OpenRouter) the model is available without regional restrictions in the documentation.

How did Muse Image perform in your test?

It scored an average of 8.9/10 on five prompts, more than GPT Image 2.5 Sunburst in “max” quality (8.8) and MAI-Image-2.6 (8.5). It wrote the sign “PIEKARNIA ŻÓŁĆ“ without errors, and the image cost $0.01.

What rank does Muse Image have?

On September 27, 2026, seventh in Artificial Analysis, with 1111 Elo. Higher were GPT Image 2.5 Sunburst and Flare, GPT Image 2, Grok Imagine Image 2.0, MAI-Image-2.6, and Nano Banana 2.

How does Muse Image differ from the Meta Muse app?

Muse Image is an image generation model in Meta AI. Meta Muse is a separate app with an AI assistant, launched September 8, 2026, only in the US.

Google · Your sourcesQuantslant higher in your Google - one clickAdd to preferred sources →

Sources and verification date

Kinetyra

// read next

More on this from QUANTSLANT

Piotr Olszewski

Editor and publisher · Quantslant

I edit and publish Quantslant. AI tools help with research, drafts and translation; choosing topics, running the tests, checking facts and the decision to publish are mine, and so is the responsibility for what appears here.

read nextMeta Muse Spark 1.3: Benchmarks, Pricing and Test ResultsbenchmarksMeta Muse Spark 1.3: Benchmarks, Pricing and Test ResultsMeta's Muse Spark 1.3 scores 48 points in the AA index, generates 214-252 tokens/s, and costs $1.25/$4.25 per million tokens.

Most read

  1. 1Best AI Video Generator Apps for Your Phone (2026)
  2. 2Claude Opus 5.5 Motion Graphics: 60s Showreel from One Prompt
  3. 3Wan 2.2 on RTX 4090: timings, VRAM and four test clips
// site map

This page in other languages2

Quantslant in numbers 1,423posts in archive22news from the last 7 days18video models in the observatory
🎬 AI Video Benchmarks: Higgsfield DoP vs OpenAI Sora • Higgsfield vs Runway Gen-3 • AI Video Credit Calculator
× ‹ enlarge ›