SHARE

a16z Just Ranked the 100 Most Popular AI Apps. Here's the Full List (And What It Tells Us).

The sixth edition of a16z's Top 100 Gen AI Consumer Apps report is out, and it reveals a consumer AI market that's splintering in fascinating ways. Here's our breakdown, plus every app ranked. Did your fave make the cut?

Written By

Grant Harvey

Mar 13, 2026

10 minute read

The app you use to edit photos, write emails, and plan your week probably has AI baked into it now. You might not even notice.

That quiet shift is exactly why a16z's sixth edition of its Top 100 Gen AI Consumer Apps report matters more than any previous version. For the first time, the list includes products where AI has become the core experience, even if they didn't start that way. Canva, CapCut, Notion, Grammarly; they all made the cut alongside the usual AI-native suspects.

First up, the TL;DR
The Race for the Default AI
The World Is Splitting Into Three AI Ecosystems
Creative Tools: Bundling Kills Standalone, Except Where It Doesn't
Agents Have Arrived (But Mostly for Technical Users)
AI Beyond the Browser
What to Watch Next
The Full List: Top 50 Gen AI Web Products (by Unique Monthly Visits)
The Full List: Top 50 Gen AI Mobile Apps (by Monthly Active Users)

First up, the TL;DR

a16z just dropped the sixth edition of its Top 100 Gen AI Consumer Apps report, and the big takeaway is that the "AI app" category barely exists anymore. AI is just... apps now.

ChatGPT still dominates (2.7x bigger than Gemini on web, nearly 30x bigger than Claude), but the three platforms are specializing fast.
- ChatGPT is going consumer super-app with shopping, travel, and transaction categories.
- Claude is doubling down on professional power users with developer tools and premium data connectors.
- Gemini is riding creative tools like Nano Banana and Veo 3 to grow its user base.
The report also tracks a splintering global market.
- Russia and China have built parallel AI ecosystems out of necessity (sanctions and censorship, respectively).
- Sber's GigaChat debuted on the web list.
- And on a per capita basis? Singapore, not the US, leads global AI adoption. America ranks 20th (probably because of the big swath of anti-AI haters out there).
Creative tools are consolidating around the big platforms for image generation, but music (Suno) and voice (ElevenLabs) remain defensible standalone categories.
Agents are the newest entrant; OpenClaw became the most-starred project on GitHub, surpassing React and Linux, before the founder got hired to work at OpenAI and spun it out into a foundation.

The bottom line: if you're only tracking ChatGPT, or your company is stuck with Copilot or Google Workspace only, you're missing how AI is actually being used. The action has moved to browsers, desktop apps, voice tools, and features embedded in software people already use.

Let's dive into all that in more detail.

The Race for the Default AI

Here's a number that should make every AI company nervous: only about 10% of the global population uses ChatGPT weekly. The market leader hasn't even scratched the surface of the total addressable market.

ChatGPT's ~900M weekly active users make it the undisputed champion. But the competitors shipped last year. Google crushed its creative models; Nano Banana generated 200M images and brought 10M new users to Gemini in its first week (we use Nano Banana to generate all our header images now!). Anthropic doubled down on the prosumer with Cowork, Claude in Chrome, and Claude Code (which hit a $1B annualized revenue run rate in six months).

The most revealing data point: the app stores are diverging almost completely. ChatGPT and Claude both have 200+ apps in their connector ecosystems, but only 11% overlap. The shared apps are the obvious productivity stack everyone needs (Slack, Notion, Figma, Gmail). Beyond that core, the platforms split:

ChatGPT has 85+ apps across Travel, Shopping, Food, Health & Wellness, and Entertainment. Booking flights on Expedia, ordering groceries through Instacart, browsing Zillow listings. It's the most aggressive play any AI company has made to become a consumer super-app.
Claude skews toward financial data terminals (PitchBook, FactSet, Moody's), developer infrastructure (Sentry, Supabase, Snowflake), and science tools (PubMed, Benchling), plus a growing open-source MCP community.
Gemini is carving its own niche through creative tools, with traction nearly perfectly correlated to model releases like Veo 3, Nano Banana 1 and 2. Google's also integrating AI across Workspace, but that growth is captured by existing products rather than a new experience.

As a16z partner Olivia Moore put it: context and memory compound. The more you use one AI assistant, the harder it becomes to switch (this is a real issue; skills make this somewhat less true, but you can't even access Skills in ChatGPT yet). Once you've configured your AI to talk to your calendar, email, and CRM, switching costs rise dramatically.

Sam Altman has hinted at a "Sign in with ChatGPT" identity layer. The idea: users carry their memory and tokens to third-party apps, effectively making ChatGPT the starting point for everything. If that plays out, this race may look less like the search wars (one winner takes 90%) and more like the mobile OS wars. Two platforms with very different philosophies, both building trillion-dollar ecosystems.

The World Is Splitting Into Three AI Ecosystems

The Western AI tools share a remarkably similar user base. ChatGPT, Claude, Gemini, and Perplexity all draw their top markets from the same pool: the US, India, Brazil, the UK, and Indonesia.

Then there's the rest of the world.

China has the lowest combined ChatGPT and Gemini usage of any country at just 15%. The market is dominated by homegrown products: Bytedance's Doubao, DeepSeek, and Kimi (we like Kimi here in the US, too).

Russia has emerged as a third distinct AI ecosystem. Sanctions created the gap, and local products filled it within two years. Yandex Browser, with its integrated Alice AI assistant, hit 71M monthly active users, making it a top-10 mobile AI product globally. Sber's GigaChat debuted on the web list. And Russia is the #2 market for DeepSeek after China (we've also seen an uptick in traffic from Russia lately; Привет, русские читатели! We think we have OpenClaw to thank for that, as OpenClaw mania has taken China, who we've also seen reading our stuff more.. 读者朋友们，你们好 !)

The per capita adoption data might be the most surprising finding in the whole report. a16z built an index combining web visits per capita and mobile MAU per capita across the products on the list:

#1: Singapore
#2: UAE
#3: Hong Kong
#4: South Korea
#20: United States
Sub-50: Russia and China

Why is the US so far down despite producing most of these products? Moore points to two factors. The US has a huge chunk of jobs where AI hasn't really touched yet (retail, transportation). And cultural trust in AI is lower here. An Edelman survey found US trust in AI sits around 32%. Top-adopting countries range from 50-80%.

Creative Tools: Bundling Kills Standalone, Except Where It Doesn't

Three years ago, seven of nine creative tools on the web list were image generators. Today, only three remain. The difference? Video, music, and voice products took the open slots.

The pattern is clear: wherever the model giants have focused (image generation, increasingly video), standalone traffic compresses. Midjourney ranked in the top 10 on the very first edition. It's fallen to #46. The products that survive have opinionated aesthetics or sophisticated workflows that ChatGPT and Gemini can't replicate as a checkbox feature.

But where the big platforms haven't invested heavily, there's room to build. Suno (#15 on the web list) held its rank. ElevenLabs has appeared on every edition since September 2023. Their capabilities (voice cloning, dubbing, audio production) remain specialized enough to resist bundling.

Video generation saw the most movement this edition. Kling AI and PixVerse have built real traction, with Chinese-developed models consistently leading in output quality. On the flip side, Sora had a massive launch (number one on the US App Store for 20 consecutive days, hit 1M users faster than ChatGPT itself) but hasn't sustained viral growth. SensorTower still counts 3M+ daily active Sora users, so the tool is sticky; it just hasn't cracked the social-app formula.

Nobody has cracked AI meets social yet. The emotional stakes just feel lower when everything is AI-generated, according to Moore. We agree: we should use AI in the background of social media apps to help us real people do things. Kinda like OpenAI's group chat experience, except instead of AI responding to every. single. message, it should shut up and only help us when we need something :) No offense to any OpenClaws out there reading this...

Agents Have Arrived (But Mostly for Technical Users)

The shift toward agentic AI started last edition with vibe coding. Lovable and Cursor appeared on the March 2025 list. This edition, the category has expanded.

The biggest story is OpenClaw, an open-source AI agent built by Austrian developer Peter Steinberger. It connects to messaging apps and executes multi-step tasks on your behalf. OpenClaw went from a solo developer's side project to 68,000 GitHub stars and mainstream coverage in weeks. As of March 2026, it's the most-starred project on GitHub of all time, surpassing both React and Linux.

The catch: OpenClaw requires Terminal knowledge to set up. It's huge among developers, but hasn't yet escaped containment to non-technical users (except in China, where people are both paying companies to install it, and then paying the very same companies to uninstall it after they realize how vulnerable it is). Apparently, visits to the setup/signup page have flatlined since early February. OpenAI acqui-hired (kinda?) the project creator in February 2026, which could signal a more consumer-friendly version is coming.

Other horizontal agents on the list include Manus (acquired by Meta for ~$2B in December 2025) and Genspark (debuted this edition with a $300M Series B and a $100M revenue run rate). Both let users hand over open-ended tasks and let AI handle the workflow end to end. One that hasn't taken off with consumers yet, but that's great for business use cases is Tasklet. The problem is, when something gets useful, the labs eventually make a similar version, and so far, Cowork has been eating up all the attention that would otherwise go to them, as best we can tell anyway. Who knows though; it's a great tool if you want a tool neutral automation platform.

These products will compete directly with the agentic capabilities the major assistants are building. Moore noted that horizontal consumer AI apps face a tough road because the big companies have both the scope and the distribution advantage. Vertical agents might have a better shot at durability. Hard agree.

AI Beyond the Browser

Every prior edition of this report (there's been five before now; in descending order, 5, 4, 3, 2 doesn't exist on Google search so basically its dead to me, and the OG!) ranked products by web visits and mobile MAU. But the most significant consumer AI growth of the past year happened in products these metrics can't capture.

Claude Code hit a $1B annualized revenue run rate in six months. OpenAI shipped Codex for Mac with 2M weekly active users as of early March. Browsers are becoming AI products: OpenAI launched Atlas, Perplexity shipped Comet, and Anthropic released Claude in Chrome. Desktop notetakers like Fireflies, Fathom, Otter, and Granola have a combined 20M web visitors across the top five players.

And AI is embedding directly into the tools people already use. Anthropic put Claude in Excel and PowerPoint. Google launched Personal Intelligence in January 2026, connecting Gemini to Gmail, Google Photos, YouTube, and Search.

The implication: a developer who spends eight hours a day in Claude Code and a knowledge worker who dictates every email through Wispr are heavy AI users who barely register in web traffic data. As AI shifts from a destination to a feature, even this list will need to evolve.

What to Watch Next

Moore identified two big trends for the next six months: voice and memory. Voice dictation is becoming normalized in tech companies and will spread to mainstream consumers within 6-9 months. And memory could become the single biggest competitive advantage in AI. As Moore put it: "Any product that you start to use two years from now, if it doesn't immediately feel like it knows you, it will feel broken."

Our take: When it comes to typing, you could almost look at it as an antiquated user interface. Not yet, and not for all, but in some ways that are very real and will manifest in the next few years, the future is likely to be a mixture of voice commands and physical manipulation of digital objects; getting us back to basics, so to speak.

There are still two places where typing will remain relevant: helping you organize your thoughts, and when you don't want to speak something publicly (so you write it down). In a lot of ways, it will be like how we stopped communicating to each other through physical handwriting; we still use it to jot down notes, or share our thoughts privately, and of course sign our name on things, but we don't send handwritten letters or contracts.

Another cool idea: the concept of onboarding to a product might not exist in a few years. And whoever owns that memory layer, whether it's ChatGPT lending it to third-party apps or users carrying it between platforms, will have built the deepest moat in consumer tech. It's very, very important that this layer is private to the user, and considered privileged in some way; Solid's data pods are a good idea around this, as is Proton's entire business model and NEAR's AI cloud is another attempt at making this a reality; we'll see who can pull it off).

The Full List: Top 50 Gen AI Web Products (by Unique Monthly Visits)

Source of the data to back this up: a16z from SimilarWeb, January 2026

The Full List: Top 50 Gen AI Mobile Apps (by Monthly Active Users)

Source of the data to back this up: a16z from Sensor Tower, January 2026

Note: Some apps on the list are character chat or companion platforms we've opted not to link to directly; sorry we're a work safe platform! The full, unedited list is available in the original a16z report.