Meta's personal AI agent Muse has surpassed 3.4 million downloads since its September 8 launch, up from 2.5 million earlier in the week. The surge follows promotion at Meta's Connect developer conference and strong early reviews. Meta has now opened an early access program for upcoming Muse features.
SynthePulse Insight · AI deep reading
Version 1 · 1 source
Meta's personal AI agent Muse shot to the top of two major U.S. app stores after launching on September 8, and opened early access to new features after the Connect conference. But download estimates from three market intelligence firms range from about 2.3 million to 4.3 million, leaving the true scale undetermined.
There is no single answer to how many downloads Meta's personal AI agent Muse has racked up since launching on September 8. Sensor Tower estimates it has surpassed 3.4 million, up from 2.5 million earlier this week; Apptopia puts total downloads at 4.3 million, including 2.6 million on iOS; and Appfigures estimates about 2.3 million, split almost evenly between iOS and Android.
The three firms' methodologies and results differ markedly, with the highest and lowest estimates nearly a factor of two apart. That means any single conclusion about the scale of Muse's growth lacks a solid foundation. What can be confirmed is that Muse is currently available only in the U.S. and Canada, and its growth figures come mainly from third-party market intelligence estimates rather than official Meta disclosure.
Beyond the absolute download numbers, Sensor Tower offers a set of relative growth figures: average daily downloads for Muse grew 55% over its first ten days, compared with 24% for ChatGPT in the same period, while Claude and Grok saw download growth decline during their respective launch periods.
This comparison comes from a single source, Sensor Tower, and reflects that firm's methodology; it cannot yet be treated as an industry consensus. But it at least points in one direction: Muse's early download ramp was faster than that of comparable conversational AI products at the same stage.
On app store rankings, Sensor Tower says Muse hit No. 1 on the U.S. App Store on September 18, topped Google Play the next day, and has held both positions since. This ranking performance is consistent with the direction of the download estimates, but it too comes from third-party observation.
Meta promoted Muse at its Connect developer conference. Sensor Tower says Muse's daily active users rose 27% the day after the conference.
But the time window matters: Sensor Tower's data may lag actual adoption and does not yet fully reflect the news announced at Connect, so its download estimate may understate the true scale. This is an inference based on the firm's own statements about data lag, not a confirmed correction.
In other words, the 3.4 million figure could be too high (relative to Appfigures' estimate) or too low (relative to actual post-conference growth). Without official Meta data, the direction and magnitude of the deviation cannot be determined.
Where Muse's growth momentum comes from — the source offers an indirect clue. Sensor Tower says Muse received the majority of Meta's internal ad placements within ten days, ahead of Facebook, Instagram, and WhatsApp, and Meta also ran ads on Reddit, TikTok, and YouTube.
But as of September 22, Muse had entered the top ten brands by ad spend, while as of September 19 ads accounted for only 6% of its impressions. From this it can be inferred that Muse's rapid growth may have been driven largely by organic traffic rather than paid promotion.
The limitation of this inference: the ad impression share is only an indirect indicator, the source does not give specific proportions for organic versus paid contributions, and the ad data only covers through September 19, a limited time window.
Meta provided Muse with the same kind of cross-platform promotional support that helped Threads surpass 500 million monthly active users. Threads head Connor Hayes told Business Insider that Muse's promotion works differently, showing the specific tasks the agent can perform for each user based on Meta's judgment of user needs.
This comparison shows Meta places Muse in the same tier of promotional resources as Threads, but the difference in approach also means Threads' growth path cannot simply be applied to predict Muse. The relevant statements come from Hayes' remarks to the media and are source claims.
Meta has opened an early access program for upcoming Muse features, inviting users to have the agent add them to the list via a prompt shared on X. The program is not a random test but a targeted invitation to AI enthusiasts who use multiple agents and understand the market.
Features announced at the Connect conference include a digital avatar that can video chat with users, expanded shopping partnerships and connectors, computer operation on the Mac app, a dedicated email address, and integration with Meta AI glasses via a wake word.
However, the source does not specify when or how broadly the early access features will roll out, so it is impossible to judge when these features will become available to a wider set of users.
All download, growth rate, and ranking figures in this article come from estimates or source claims by third-party market intelligence firms (Sensor Tower, Apptopia, Appfigures); Meta has not officially disclosed the relevant numbers. The three firms' estimates differ markedly, with the highest and lowest nearly a factor of two apart, and Sensor Tower itself says its data may lag actual adoption. The time windows for ad-related data are limited (impression share through September 19, ad spend ranking through September 22). Content marked as inference in the article (such as organic-traffic-driven growth or data understatement) is reasonable extrapolation based on source facts, not confirmed conclusions.
The Muse download dispute is itself more informative than any single number: it shows that, absent official Meta data, the divergence among third-party estimates is enough to rob the question of "how fast is it growing" of a definitive answer. What can be confirmed is the No. 1 ranking, the low ad impression share, and the rise in activity after the Connect conference; whether the true scale is 2.3 million or 4.3 million still awaits a more complete methodology or official disclosure.
Primary report
Primary source