Muse AI Agent Beats ChatGPT to No. 1 in the US App Store
Meta's Muse AI agent took the top spot on Apple's US App Store free chart roughly a week after its September 8 debut, pushing OpenAI's ChatGPT into second place. Alexandr Wang, Meta's chief AI officer, announced the ranking. Meta describes Muse as a personal agent that completes tasks rather than answering questions: it books travel, sends email and fills in forms on the user's behalf.
Sensor Tower estimates put Muse past 730,000 US downloads by the September 18 chart capture. The app is limited to US users for now and runs on iPhone, Android, the web and inside WhatsApp. On the same free iPhone chart it also ranks above Instagram, Gemini, Claude, Grok and Copilot.
The No. 1 position is the news. Its meaning is narrower than it looks. A chart rank counts downloads, and a download is the cheapest thing a user can hand an agent. The question the ranking raises, whether consumer demand is moving from chatbots that answer to agents that act, is settled by whether those users grant an agent the access it needs to act at all.
What the Ranking Measures
App Store charts rank install velocity, not engagement. First place means more people downloaded Muse than any other free iPhone app during the window measured. It says nothing about how many of them opened the app twice, how many tasks the agent completed, or how many downloads became paid subscriptions. Meta attributes the climb to a wave of new installs and a rapid sequence of feature rollouts, both of which the chart rewards directly.
The growth curve was steep, then flattened. Muse cleared 83,000 US iOS downloads by September 10, enough for second place at the time, and passed 730,000 by September 18. Sensor Tower's estimates place it near 600,000 across the first five days, so the daily rate fell sharply in the second half of that stretch. The ranking reflects the size of the accumulated base, not a surge on the day it was recorded.
| App | Rank, US free iPhone chart |
|---|---|
| Meta Muse | 1 |
| ChatGPT | 2 |
| Google Gemini | 8 |
| Anthropic Claude | 13 |
| Grok | 32 |
| Microsoft Copilot | 64 |
The gap at the top is one slot. Below it the field spreads out, with the four other major assistants sitting between eighth and 64th. Consumer attention is concentrating on a small set of brands instead of spreading across every model provider, and Muse is the newest entrant in that group.
The download figures come from Sensor Tower's modelling rather than from Meta. Meta's announcement centred on the chart position, which leaves the scale of the launch resting on third-party estimates of iOS installs in a single market. Neither company has released daily active usage numbers, so the comparison stops at installs.
The Shift From Answering to Doing
The distinction the Muse AI agent is built on is where the work stops. A chatbot returns an itinerary, a draft or a form with placeholder text, and the user finishes the job. An agent is meant to complete the booking, send the message or submit the form, moving the user from writing prompts to approving actions.
That change is larger than it sounds. Chatbots are cheap to try and cheap to abandon, because a bad answer costs a re-prompt. An agent's mistakes carry the cost of the underlying task, since a wrong flight or an unintended email cannot be undone by ignoring a reply. Trust has to be extended before the first result rather than after it, which is why agent adoption should lag install numbers instead of tracking them.
The two formats also suit different jobs. Chatbots remain the better tool for open-ended work where the user is still deciding what they want: research, drafting, comparison, a first pass at a plan. Agents fit tasks with a defined end state and no interesting decisions in the middle, such as rebooking a standard flight, sending a routine reply, or filling in a form the user has filled in before. Muse's feature set of travel booking, email and forms sits squarely in that second category, where the demand case for agents is strongest and also narrowest.
Meta is betting the second category is bigger than the first. The chart position shows appetite for trying that bet. It does not show that users prefer the agent to the chatbot for the same job.
Access, Retention and the Limits of the Claim
An agent needs something a chatbot does not: access. Booking travel, sending email and completing forms means Muse touches a user's inbox, calendar, payment methods and travel accounts. Whether people leave that access switched on once the novelty fades is the question the chart cannot answer, and it decides whether the launch carries commercial weight beyond visibility. An agent holding a wallet and a booking history sits closer to a transaction than a chat window does.
Permissions cut both ways for Meta. A single inbox or payment connection is what turns a general assistant into something that can finish a real errand, and it also concentrates risk in one place if the agent misfires on a booking or a send. That trade-off sets how quickly users extend the agent's reach beyond the first task they try.
Geography narrows the claim further. Muse is a US-only release, so first place covers one country on one operating system. Meta's distribution advantage comes from shipping the agent inside WhatsApp as well as standalone apps, a surface the pure AI labs do not control, but that reach counts only once the rollout moves past the United States.
How the Launch Compares With Meta's Earlier Apps
| Launch | Early US downloads |
|---|---|
| Threads | 4.3 million+ on day one |
| Meta AI | 108,000 at debut |
| Muse | 83,000 by September 10; 730,000+ by September 18 |
| ChatGPT | 500,000+ within its first week |
Against Meta's own record, Muse is a mid-sized launch. Threads drew more than 4.3 million US downloads on its first day, and the Meta AI app recorded 108,000 at debut. Muse's roughly 600,000 installs across its first five days land well below Threads and above Meta AI's opening figure.
The ChatGPT comparison is closer than the chart position implies. ChatGPT crossed half a million US installs inside a week, a pace Muse has broadly matched. Beating a four-year-old app in a download ranking reflects the difference between a fresh launch surge and an established install base, not a verdict on either product.
Why this matters
If the install spike becomes habitual delegation, Meta gains a consumer entry point closer to transactions than any of its social apps, and the chatbot stops being the default shape of consumer AI. If users switch access off after the launch window, the No. 1 ranking reads as a download event rather than a change in behaviour. What matters next is how many of those 730,000 downloads stay active.
Related Articles
- Meta's Muse AI agent sends emails, books trips and shops for $20 a month
- Muse Spark 1.3: Meta's most powerful model arrives with a paid-API pitch
- Meta Muse Image Brings Deliberate Reasoning to AI Photo Generation Across Instagram and WhatsApp
✔Human Verified
Researched and cross-referenced against primary sources by the Bytevyte editorial team. This article was generated with the assistance of artificial intelligence and reviewed by the Bytevyte editorial team.