Meta's latest image and video AI tools enable you to transform Instagram into your personal creative mood board.
Meta has been developing something exciting, and today, it's finally unveiled it. On July 7, 2026, Meta Superintelligence Labs introduced Muse Image and Muse Video (in preview), its first internal media generation models.
The launch features several aspects that are difficult to dispute.
Presenting Muse Image and Muse Video, the initial media generation models crafted by Meta Superintelligence Labs. Muse Image stands as our most sophisticated image generation model yet. It accurately follows instructions, edits precisely, creates from various references, and produces… pic.twitter.com/byNpQZO1RW— AI at Meta (@AIatMeta) July 7, 2026
What truly sets Muse Image apart?
You can directly mention a public Instagram account (with ‘@’) in your prompt, allowing Muse Image to use images from that profile as visual references for your design.
This means if you desire a personalized event invitation that reflects your brand's style, or if you're collaborating on a creative idea, you can easily tag the relevant account.
This is an innovative feature that links image generation directly to the social graph that Meta has led for twenty years.
Moreover, Muse Image operates more like an agent compared to a conventional text-to-image model.
Great news: Meta's Muse Image has secured the #2 position in the Image Arena! Muse Image from @AIatMeta now ranks just behind OpenAI’s GPT Image 2, surpassing Nano Banana, Grok Imagine, MAI Image, and several other top image models. It holds the #2 spot universally: Text-to-Image,… https://t.co/46ugzuajF7 pic.twitter.com/577214iiqj— Arena.ai (@arena) July 7, 2026
How does it really compare to its competitors?
Instead of producing a single outcome based on your prompt, it evaluates the request, looks up factual context online, writes and executes code when accuracy is crucial (for charts or QR codes), and self-corrects its output before providing you with the final result.
On Arena.ai’s human-preference leaderboard, which ranks models based on actual user votes, Muse Image occupies the second position across text-to-image, single-image editing, and multi-image editing as of July 5, 2026.
The only model ranking higher is OpenAI’s GPT Image 2, with a score of 1,385, compared to Muse Image’s 1,280. Meanwhile, Google’s Nano Banana 2 (1,270) and MAI Image 2.5 (1,257) follow behind.
Muse Image is available for general use in the Meta AI app (meta.ai), Instagram Stories (in the US), and WhatsApp in select countries. Facebook support will be added soon. For those needing higher-volume creation, a subscription option is available.
1/ Today, with the launch of Muse Image and the preview of Muse Video, we wanted to share additional samples from the model along with research details and evaluation results. See more samples from Muse Video below (and check the thread for the research blog!) pic.twitter.com/7yVUOolkG4— Alexandr Wang (@alexandr_wang) July 7, 2026
What about Muse Video?
Built on the same foundational training as Muse Image, Muse Video can generate native audio, producing sound alongside visuals.
On Arena.ai, it has already achieved the #3 ranking for text-to-video in human preference. Meta has been transparent about the need for improvements in audio-video synchronization and fast motion physics, which is why this is currently a preview rather than a full launch.
Muse Video will be available soon for creators and Meta AI, with no specific release date yet.
To me, Muse Image and Muse Video represent more than just product launches. They signify Meta’s departure from relying on third-party AI providers, which it has used for years. Developing in-house allows Meta to iterate more quickly, integrate seamlessly, and exert greater control over the advertising tools it is increasingly monetizing.
Other articles
The arrest of a hacker has disclosed how Microsoft is able to monitor your Windows device.
A hacker utilized a VPN to maintain anonymity; however, Microsoft's Windows Global Device Identifier recorded his actions and reported them to the FBI.
In your upcoming Teams meeting, there might be an AI assistant that responds to questions on your behalf.
Microsoft's summer Teams update is more utilitarian than others, featuring AI-enabled notes in in-person meeting spaces and a chat list that autonomously arranges itself.
Solos' newest AI glasses are introducing a true privacy mode.
Solos has introduced the camera-less AirGo A6 smart glasses together with a Privacy Kit for the AirGo V2 that physically obstructs its integrated camera.
Sources indicate that DeepSeek has initiated the development of its own AI chips to reduce dependence on NVIDIA.
In recent years, as generative AI has evolved from model training to large-scale inference, more AI model developers have shifted their focus.
A bug in Windows 11 could be silently consuming hundreds of gigabytes of your storage space.
If system files are consuming hundreds of gigabytes on your Windows 11 computer, a bug related to CapabilityAccessManager.db-wal might be the cause.
You can now install Windows on the Steam Machine, for reasons only you comprehend.
Valve has launched official Windows drivers for the Steam Machine, simplifying the process of installing Microsoft's operating system, although SteamOS still seems to provide superior gaming performance.
Meta's latest image and video AI tools enable you to transform Instagram into your personal creative mood board.
With Muse Image now live and Muse Video in preview, Meta has officially ended its collaboration with Midjourney and Black Forest Labs for its creative AI.