Post-Cutoff
Search For AI agents

MetaMuse familyCurrentImage generationReleased 7 Jul 2026

Muse Image

Meta Superintelligence Labs’ first image model: an agent that searches, codes and self-checks before returning an image.

Model idmuse-image-1.0

Price$0.01 per image

Checked25 October 2026

Where it stands

12newer image generation models since its release

The 5 newest are listed. All image generation models

Weights
Closed weights; access is hosted only.

How to call it

5 ways to use Muse Image, with the model id each provider expects.

Providers, model ids, endpoints and docs for Muse Image
ProviderModel idEndpointDocs
Meta Model API muse-image-1.0 https://api.meta.ai/v1/images/generations dev.meta.ai
Meta Model API (edits) muse-image-1.0 https://api.meta.ai/v1/images/edits dev.meta.ai
fal fal-ai/meta/muse-image fal.ai
OpenRouter meta/muse-image openrouter.ai
Web app meta.ai

Pricing

Prices for Muse Image
Per image$0.01

As published: “USD per generated image (Meta Model API; search grounding included)”.

Source: dev.meta.ai, checked 25 October 2026.

Specs

Input
Text and images
Output
Images
Context window
Not published
Max output
Not published
Open weights
No
Licence
Proprietary
Released
7 July 2026
Status
Current
Type
Image generation

What stands out

What its maker marketed at launch, and what people found later, each with its source.

  • Agentic image generation

    Plans, searches the web for references, writes and runs code where precision matters (plots, QR codes) and checks its result; self-refinement (local edit, regenerate or switch tactics) emerged during RL training.

    Source: ai.meta.com

  • Test-time compute scaling

    Meta reports quality rising roughly log-linearly with inference-time compute; ComfyUI exposes reasoning strength high/low and toggles for web search, image search and code execution.

    Source: ai.meta.com

  • Multi-reference composition and multi-turn editing

    Combines people, objects, clothing, styles and environments from several inputs with interleaved text and images; multi-turn refinement through the Responses API.

    Source: dev.meta.ai

  • One-cent pricing

    Found after launch

    $0.01 per image on the Meta Model API with search grounding included (Artificial Analysis lists $10 per 1k images).

    Source: dev.meta.ai

Notes

Artificial Analysis text-to-image leaderboard (read 2026-10-25): #8, Elo 1127, listed release Jul 2026, $10.0 per 1k images (https://artificialanalysis.ai/text-to-image/arena/leaderboard-text). AA image-editing: tied 7-14, Elo 1129. Consumer launch 2026-07-07 (Meta AI app, meta.ai, Instagram Stories US, WhatsApp in some countries). API launch reported 2026-08-28 (RuntimeWire, Alexandr Wang X post https://x.com/alexandr_wang/status/2093390371875414501); OpenRouter lists it from 2026-08-26 and its endpoints list was empty when checked, so OpenRouter availability is unconfirmed. Output about 2.5 MP and 14 aspect ratios plus auto (per ComfyUI partner-node docs, not Meta). Images made in Meta’s apps carry an invisible Content Seal watermark. No dedicated launch entry in the timeline yet (entry field points to the Model API entry, which mentions it). Rate limits not published.

curl https://api.meta.ai/v1/images/generations -H "Authorization: Bearer $MODEL_API_KEY" -H "Content-Type: application/json" \
 -d '{"model":"muse-image-1.0","prompt":"An infographic of the solar system with correct planet order"}'

Sources: