Meta has been cooking something up, and today, it finally put it on the table. On July 7, 2026, Meta Superintelligence Labs launched Muse Image and Muse Video (in preview), its first in-house media generation models.
The rollout comes with a few features that are genuinely hard to argue with.
What makes Muse Image actually different?
You can mention a public Instagram account (using ‘@’) directly in your prompt, and Muse Image will use photos from that profile as visual references for your design.
So, if you want a custom event invite that features your brand’s aesthetic, or you’re building a creative concept with a collaborator, you can simply tag the required account.
It’s a genuinely novel feature, one that ties image generation directly to the social graph Meta has dominated for two decades.
Beyond that, Muse Image works more like an agent than a traditional text-to-image model.
How does it actually stack up against the competition?
Rather than generating a single output based on your prompt, it thinks through the request, searches the web for factual context, writes and runs code when accuracy matters (for charts or QR codes), and self-corrects its output before handing you the result.
On Arena.ai’s human-preference leaderboard, which ranks models based on real user votes, Muse Image holds the second spot across text-to-image, single-image editing, and multi-image editing as of July 5, 2026.
Only OpenAI’s GPT Image 2 ranks higher, at 1,385, compared to Muse Image’s 1,280. Meanwhile, Google’s Nano Banana 2 (1,270) and MAI Image 2.5 (1,257) trail behind.
Muse Image is free for everyday use in the Meta AI app (meta.ai), Instagram Stories (US), and WhatsApp in select countries. Facebook support is coming soon. For those who require higher-volume creation, a subscription tier unlocks that.
What about Muse Video?
Built on the same pretraining base as Muse Image, Muse Video supports native audio, generating sound alongside visuals.
On Arena.ai, it already ranks #3 for text-to-video in human preference. Meta has been upfront that audio-video synchronization and fast motion physics still need work, which is exactly why this is a preview and not a full launch.
Muse Video is coming soon to creators and Meta AI, with no firm date attached yet.
To me, Muse Image and Muse Video aren’t just product releases. They represent Meta cutting ties with third-party AI vendors it has depended on for years. Building in-house gives Meta faster iteration, tighter integration, and control over advertiser tools it’s increasingly monetizing.