Meta Muse Daily Users Reportedly Plateau at 2.5 to 3 Million: How to Read the Number

When Meta launched Muse on September 8, 2026, the growth chart looked vertical. Our coverage followed its climb to the top of the App Store, and the product's token-heavy early usage. Now a report dated October 5 says the line has flattened: daily active users stayed between 2.5 million and 3 million from late September through October 1, with no meaningful growth for more than a week.

Numbers about private apps are always shaky, and these are no exception. This post explains what is reported, why different sources disagree, how to read a plateau in an agent product, and what it suggests for anyone building something similar.

TL;DR: the questions people are asking

What is reported

The facts that circulated on October 5:

• Launch: September 8, 2026.
• Early growth: about 700,000 daily users by September 24, described as a tenfold rise in 11 days, with 5 million downloads reached shortly after launch.
• Plateau: 2.5 to 3 million daily active users across web, iOS and Android from late September to October 1.
• Source: brokerage data in an analyst note, without a named tracking vendor in the coverage we reviewed.
• Response: Meta has been adding integrations for small businesses, including Shopify, QuickBooks, Stripe and Canva.

Alexandr Wang, who leads Meta's AI effort, listed what Muse shipped recently on October 5: connectors, invite codes, a phone-call beta, Muse for Mac, Mac computer use, a connector platform, small-business tools and gadgets. Several of those we covered in detail, including Muse for Mac, the developer connectors platform and business-in-a-box commerce.

Why the numbers disagree

You will see different Muse figures depending on the source. That is normal, and it is worth understanding why before drawing conclusions.

An analyst cited about 1.8 million daily users around the same period, and earlier reports said Muse had not yet reached a million. Those are not necessarily contradictions, because they may be different dates or methods. The sensible approach is to track one source over time and compare the trend, not to average across sources. We also have not seen Meta publish daily active user figures for Muse, so every number here is an outside estimate.

How to read a plateau in an agent product

A flat curve can mean very different things.

A healthy plateau. The product retained most of the people it acquired, and new-user flow slowed after the launch spike. The base is stable and can grow with new features. This is common after a viral launch.

A leaky plateau. New users still arrive, but others leave at a similar rate, so the total stays flat. The total hides churn. This is a warning.

A capacity plateau. Growth is limited by supply, such as invite codes or compute. Muse used invite codes and earlier reporting noted service strain, so this is plausible, though we cannot confirm it from the data. Our coverage of the invite-code rollout discusses the constraint.

A measurement plateau. The estimating firm's panel stopped capturing growth. Third-party data can plateau artificially.

The data we have cannot distinguish these. Retention cohorts, meaning the share of day-one users still active on day 7 and day 30, would, and they are not public.

What agent products need to earn daily use

A chat app earns daily use through habit: you open it when you have a question. An agent product has a harder job, because it competes with your time-saving alternatives and needs recurring tasks to justify opening. A few forces matter.

• A recurring job. Agents that handle something weekly or daily, like inbox triage, scheduling or shopping lists, build habit. One-off tasks do not.
• Reliability. If a task works on Monday and fails on Wednesday, trust drops fast. We discussed the fragility of agents that depend on websites in our opinion piece on agent teams, and Amazon's block on Muse shopping is a concrete case.
• Trust. Privacy and permission incidents, such as the Apple Full Disk Access dispute, affect willingness to grant access.
• Distribution. Meta can promote Muse across its apps, which helps acquisition more than retention.

Meta's pivot toward small-business workflows is consistent with this logic: business tasks recur, have clear value and can justify daily use.

What builders should track instead of headline users

If you build an agent product, three metrics tell you more than total users.

1. Retention by cohort. Day-7 and day-30 return rates for users who completed a first task.
2. Delegation depth. The share of users who give a second, third and tenth task, and how many tasks per active user per week.
3. Task success and recovery. The share of tasks completed without human rescue, and how often failures cause churn.

Also watch cost per active user. Reports of heavy token use per user in the early days suggest agents can be expensive to serve, which makes retention economics important. A product with flat users and rising cost has a different problem than one with flat users and falling cost.

A simple retention model

To see why a plateau is ambiguous, imagine two scenarios with the same flat total of 2.7 million daily users. In the first, 100,000 new people start each day and almost nobody leaves, so the base would be growing, which contradicts the flat line, so this is not it. In the second, 400,000 new users arrive daily while 400,000 stop using the product, a very high churn rate that nets to zero. In a third case, almost nobody arrives and almost nobody leaves, a stable but stagnant base. The reported number looks identical in the second and third cases, but they imply very different futures: the second says the product is a leaky bucket and acquisition must continue forever, the third says it has a loyal niche.

That is why outside analysts focus on cohorts when they can get them. Only Meta knows the real shape. For now, the plateau should change your view of Muse's momentum modestly, and should not change your view of its long-term potential, which depends on whether business integrations give people reasons to return daily.

Comparing launches fairly

When people compare Muse with other agent launches, they often compare peak numbers measured at different points in the product's life. A fair comparison uses the same time since launch, the same metric (daily rather than weekly users) and the same source. Without that, a viral first week looks like a failure next to a slow burn, or the reverse. Treat any single cross-product ranking with care.

What this means for what you build or pay

For consumers, nothing here changes whether Muse is useful to you. For builders, the lesson is to treat spike metrics with suspicion, including your own. A launch can produce impressive numbers that say little about habit. And for anyone evaluating Meta's position against rivals such as OpenAI's Dots and SpaceXAI's Grok Bot, a fair comparison needs retention and usage-depth data, which none of them publish. Our five-agent comparison focuses on capabilities and risk for that reason.

The honest summary is that a reported plateau at 2.5 to 3 million daily users is a modest warning sign, not a verdict. We will update this post if Meta or a named analytics vendor publishes better data.

Related reading

• Meta Muse hits number one on the App Store
• Muse: one billion tokens per user and invite codes
• Muse for Mac
• Muse developer connectors platform
• Is Meta's Muse safe to use?
• What Meta got right with Muse
• Agent teams are the new org chart
• Dots vs Grok Bot vs Muse vs OpenClaw vs Hermes

Primary: HuggingNews summary of the analyst note (October 5, 2026) · public posts by Alexandr Wang · related reporting on Muse usage

Figures are third-party estimates as of October 6, 2026, from a brokerage note whose underlying data source was not named. Meta has not confirmed daily active user counts. Treat them as directional.