Meta’s Muse AI hits 5 million downloads in 22 days, outpacing ChatGPT, Grok and Claude
Meta’s Muse AI assistant has blown past 5 million U.S. downloads just 22 days after launch, reaching the milestone far faster than ChatGPT, Grok or Claude and giving Meta an early lead in the race to put autonomous AI agents into consumers’ hands.
Sensor Tower estimates Muse crossed 5 million downloads after launching Sept. 8. ChatGPT needed 56 days to reach the same U.S. milestone, Grok took 103 days, and Claude took 492 days, according to the market intelligence firm. Muse remains at the top of the U.S. Apple App Store charts.
That puts Muse in rare company. Sensor Tower says just 23 apps have surpassed 5 million U.S. downloads within 22 days of launch. Roughly 85% were mobile games such as Fortnite, Pokémon GO and Candy Crush. Disney+, HBO Max and Meta’s own Threads were among the few non-gaming apps to get there that quickly.

Meta Muse at No. 1 of App Store (https://apps.apple.com/us/iphone/charts/36?chart=top-free)
Muse is different from the chatbot Meta spent years trying to make competitive with ChatGPT. Powered by the company’s Muse Spark models, the app is built as a personal agent that can browse websites, send emails, fill out forms, book travel, and work across connected services on a user’s behalf. It runs inside a dedicated virtual machine in Meta’s cloud and can remain working after a user hands it a task.
Meta has something its AI rivals don’t
Muse’s numbers come with a giant asterisk: Meta owns one of the biggest distribution machines on the planet.
Sensor Tower found that Meta began promoting Muse through house ads shortly after launch. By late September, Muse was receiving as much as half of Meta’s daily house-ad impressions, taking promotional inventory that had previously gone to Facebook and WhatsApp. Meta also pushed Muse outside its own apps, buying ads across Reddit, TikTok, YouTube and mobile ad networks.
The comparisons with rival AI apps aren’t perfectly symmetrical either. Muse launched on both iOS and Android in the U.S. and Canada. Grok and Claude were limited to iOS during their first five days, according to Sensor Tower. ChatGPT entered an entirely different market in 2022 as the product that introduced generative AI chatbots to much of the public.

None of that makes 5 million downloads insignificant. Meta has spent tens of billions of dollars building AI infrastructure, models, and data centers. Muse gives the company something investors have been waiting to see from that spending: a consumer AI product people are actually installing at scale.
That growth also has money attached. Muse has a free tier alongside subscriptions priced at $20 and $100 per month for heavier usage. Meta has discussed turning agents into another transaction layer, potentially collecting fees as they help users shop, book services, or interact with businesses. Truist analysts have estimated Muse could contribute $28.5 billion in incremental revenue by 2030.
Five million downloads is the easy test
The harder question is what happens after people install it.
If Muse is going to act like a digital assistant, it needs access to far more personal information than a conventional chatbot. Emails, calendars, online accounts and other data can become part of the equation. Meta says Muse runs inside its Secure VM, keeps credentials in protected storage and gives users control over connected services.
Those safeguards are already being tested.
Meta strengthened Muse’s safety warnings after a researcher uncovered a vulnerability that could have exposed sensitive information stored inside users’ virtual machines, Reuters reported. Meta classified the incident as SEV-2 on its internal severity scale. In another episode, Muse shared a YouTuber’s home address with a Facebook Marketplace buyer after the user had granted the agent broad permissions.
Amazon has thrown up another obstacle, blocking Muse from shopping on Amazon.com and citing unauthorized automated access. That dispute points to a bigger question facing every autonomous agent: an AI can act across the web only when other companies let it through the door.
Competition is arriving quickly. OpenAI launched Dots this week, an always-on agent built to pursue goals across applications, putting another heavyweight directly into the personal-agent race.
Meta has already proved it can put Muse on millions of phones. The next milestone matters more: whether those millions keep coming back once the curiosity, mascot, and advertising blitz wear off.
If they do, Muse may represent something far bigger than another successful app launch. It could mark the moment the consumer AI battle began shifting from who has the chatbot people ask questions to, to who owns the agent people trust to act for them.

