Meta Muse (Image + Video)
A Tier · 8.4/10
Meta Superintelligence Labs' first media-generation models (launched 2026-07-07) -- Muse Image is GA across the Meta AI app, Instagram, and WhatsApp (Arena #2 for text-to-image); Muse Video is in preview (Arena #3 for text-to-video). Instruction-faithful editing, multi-reference composition, tool use, and Content Seal provenance watermarking
Score Breakdown
The Good and the Bad
What we like
- +Frontier-lab quality on day one: Muse Image ranks #2 on Arena for text-to-image, single-image editing, AND multi-image editing (as of 2026-07-05) -- a strong debut for Meta's first media model
- +Instruction-faithful editing + multi-reference composition: it follows edit instructions precisely and composes from multiple reference images, the practical features that matter for campaign/brand consistency
- +Genuinely free and where people already are -- built into the Meta AI app, Instagram Stories, and WhatsApp rather than a separate paid product to learn
- +Provenance built in: images carry a hidden Content Seal watermark (a step ahead of most consumer image tools on AI-content transparency)
What could be better
- −Muse Video is preview-only at launch ('coming soon to creators and Meta AI') -- you cannot rely on it for production video yet
- −No public API or pricing disclosed -- developers can't build on it, and there's no paid tier for higher limits or commercial guarantees yet
- −Arena #2 (image) / #3 (video) means it trails the current leaders -- it's excellent, not the outright best in either category on launch day
- −Rollout is uneven and region-gated (Instagram Stories US-only, WhatsApp 'limited countries,' Facebook 'coming soon') -- availability depends heavily on where you are
Pricing
Free (in Meta apps)
- ✓Muse Image available in the Meta AI app + meta.ai, Instagram Stories (US), WhatsApp (limited countries); Facebook coming soon
- ✓Muse Video in preview -- coming soon to creators and Meta AI
- ✓No standalone pricing or public API disclosed at launch
- ✓Meta AI app usage limits apply
Known Issues
- FEATURE PULLED AFTER BACKLASH (~2026-07-10): Meta discontinued the Instagram feature that let users **generate AI images using photos from other people's public Instagram accounts**, days after it launched with the Muse rollout. The default was automatic opt-in, which drew immediate criticism from entertainment-industry figures and SAG-AFTRA ('Anything other than a clear and conspicuous opt-in for these types of uses of Instagram users' images is unacceptable'). Meta's statement: 'We've heard the feedback that this feature missed the mark, so it's no longer available.' IMPORTANT SCOPE: this kills only the public-photos remix feature -- **the Muse Image model itself remains live** in the Meta AI app, Instagram Stories, and WhatsApp. Worth knowing as a consent/provenance signal when evaluating Meta's media toolsSource: Reuters (2026-07-10), Meta statement, SAG-AFTRA statement, Benzinga, Medianama · 2026-07-10
- LAUNCH (2026-07-07): Meta Superintelligence Labs (led by Alexandr Wang) launched its first media-generation models. **Muse Image is generally available** across the Meta AI app, meta.ai, Instagram Stories (US), and WhatsApp (limited countries), coming soon to Facebook -- capabilities include instruction-faithful editing, multi-reference composition, tool use (coding + web search), self-refinement, and test-time compute scaling. It ranks #2 on Arena for text-to-image, single-image editing, and multi-image editing (as of 2026-07-05). **Muse Video is in preview** ('exceptional visual fidelity with native audio support,' Arena #3 for text-to-video), coming soon to creators and Meta AI. Images carry a hidden Content Seal provenance watermark. No pricing, benchmarks beyond Arena placement, or public API were disclosed at launchSource: Meta AI blog (ai.meta.com/blog/introducing-muse-image-muse-video-msl/), Meta newsroom (about.fb.com), Reuters, CNBC · 2026-07-07
Best for
Anyone already in Meta's apps (Instagram, WhatsApp, Meta AI) who wants free, high-quality image generation and editing with strong instruction-following -- especially for social content where the tool lives right where you post.
Not for
Developers who need an API (there isn't one yet), teams that need production video today (Muse Video is preview-only), or anyone outside the launch regions. For a public API or top-of-Arena fidelity, use Nano Banana 2 or a dedicated provider.
Our Verdict
Meta Muse is a serious debut: a frontier lab's first image model landing at #2 on Arena across text-to-image and both editing tracks, free, and embedded directly in the apps where billions of people already share photos. The instruction-following and multi-reference composition are the features that make it useful beyond novelty, and shipping Content Seal provenance from day one is the right call. The caveats are real -- Muse Video is preview-only, there's no API or pricing, and the rollout is region-gated -- so treat this as an excellent in-app consumer tool today rather than a production or developer platform. If you live in Instagram/WhatsApp, it's an easy default; if you need an API or the outright quality ceiling, look at Nano Banana 2.
Sources
- Reuters (via AOL): Meta discontinues AI image feature days after launch (2026-07-10) (accessed 2026-07-18)
- Meta AI blog: Introducing Muse Image + Muse Video (2026-07-07) (accessed 2026-07-07)
- Meta newsroom: Introducing Muse Image (accessed 2026-07-07)
- Reuters: Meta expands generative AI tools with Muse Image rollout (accessed 2026-07-07)
- CNBC: Meta AI Muse Image (accessed 2026-07-07)
Explore more Meta Muse (Image + Video) rankings
Deeper leaderboards, benchmarks, task-specific tier lists, and status/pricing pages for Meta Muse (Image + Video).
The Tier List Tuesday
Weekly newsletter: tier movers, new entrants, and the VS of the week. Built from our daily AI-tool sweeps. No spam, unsubscribe anytime.
Alternatives to Meta Muse (Image + Video)
Midjourney
**Midjourney's alpha site gained shared folders and a Thinking mode in its 10/7 changelog: collaborators can now generate into a shared folder, and 'Rerun (Thinking)' has V8.2 inspect its own image for missed prompt details -- wrong objects, layout, anatomy, text -- and generate again.** Industry-leading AI image generation with stunning artistic quality. **The V8.2 Edit Model opened to everyone on 2026-08-27** -- instruction-based edits, up to four image references (replacing omni-reference), inpainting and outpainting -- and the **2026-09-24 update** makes targeted edits change only the pixels you select, adds live style previews, and fixes --tile seams on V8.1/8.2. V8.2 itself launched 2026-07-24 as the default model. No price change
DALL-E (Shut Down)
OpenAI's DALL-E 2 and DALL-E 3 -- SHUT DOWN. Both APIs were retired on 2026-05-12. DALL-E 3 was already removed from ChatGPT in December 2025. Existing integrations now fail; migrate to gpt-image-1.5 / gpt-image-1-mini (request shape differs -- not a drop-in swap). Tier-list alternatives: Nano Banana 2, Midjourney, FLUX.2 [klein], Ideogram
Stable Diffusion
Open-source AI image generation with unlimited free local use and full customization
Leonardo AI
Versatile AI image generator with fine-tuned models and a generous free tier
Adobe Firefly
Adobe's AI image generator -- commercially safe and baked into Creative Cloud. Firefly AI Assistant public beta is now LIVE globally (2026-04-27) for Creative Cloud Pro and paid Firefly plans, orchestrating multi-step workflows across Photoshop, Lightroom, Premiere, and Illustrator from a single chat
Ideogram
AI image generator that nails text rendering -- Ideogram 4.0 (June 3 2026) goes OPEN-WEIGHT with commercial license: 9.3B DiT, 0.97 X-Omni OCR, layout control, 2K photoreal. Plus Custom Models for brand-trained generation on Pro+ plans
![Flux (FLUX 3 + FLUX.2 [klein]) logo](/logos/flux.png){kind=small}
Flux (FLUX 3 + FLUX.2 [klein])
Black Forest Labs' model family has moved past still images. **FLUX 3** is a single multimodal foundation model trained across images, video and audio (Early Access 2026-07-23); **FLUX 3 Video went GA on the BFL API on 2026-08-04** with clips up to 20 seconds, native audio, multi-shot scenes and 1080p output, priced per second of video ($0.17/s HD text-to-video). **FLUX 3 Action (2026-09-22)** is an open-weight 7B world action model for robots. The open-core FLUX.2 [klein] image models (Jan 2026, sub-0.5s generation) remain the local-inference workhorse
Krea AI
Real-time AI image generation and enhancement with a visual, interactive canvas
NightCafe
Community-driven AI art generator with multiple models, daily free credits, and a social gallery
Nano Banana 2.1 (Gemini Nano Banana 2.1, formerly Nano Banana 2)
**Nano Banana 2.1 went GA on the Gemini API on 2026-10-06 as the update to Nano Banana 2, with better visual quality, prompt adherence, multi-turn character consistency and text rendering, new 1:4 to 8:1 panoramic ratios, and a lower price -- $0.0336 per 1K image versus $0.067 -- while Google is retiring gemini-3.1-flash-image in its favour (no end date yet).** Google's Gemini 3.1 Flash Image model -- the best-in-class text-in-image renderer, now the default across the Gemini app
Microsoft MAI-Image-2.6
Microsoft's in-house image model -- **MAI-Image-2.6 launched 2026-08-10: No. 2 on Arena text-to-image (+79 Elo over 2.5, text rendering +91), and as of 2026-09-04 back to **No. 2 on Arena image editing** as well, plus No. 2 text-to-image and No. 1 image editing on Artificial Analysis -- ahead of Google's Nano Banana family and Meta's Muse Image. **On 2026-09-04 it reached Microsoft Foundry for developers alongside a new MAI-Image-2.6-Flash** (2.8x faster than GPT-Image-2-Medium at 72% greater efficiency), both adding multi-image reference editing, web grounding and dynamic aspect ratios.** The prior MAI-Image-2.5 launched 2026-06-02 at Build (#2 image editing, #3 text-to-image on Arena, surpassing Nano Banana Pro's Arena score; +75 overall vs MAI-Image-2). Live in PowerPoint + OneDrive, on Azure Foundry, MAI Playground, and OpenRouter. Ultra-efficient MAI-Image-2.5-Flash now LIVE in Foundry ($1.75/$1.75/$19.50 per 1M). Original MAI-Image-2 shipped 2026-04-02