Similarweb's April 2026 AI chatbot traffic data is out. Seven platforms. 10.07 billion combined web visits. ChatGPT at 54.7% worldwide. Gemini at 27.4%. Claude at 8.2%.

The number that should stop you: Gemini was at 5.6% in February 2025. Fifteen months ago. It gained 21 percentage points. ChatGPT fell 22 points in the same stretch.

I want to say upfront what this data is and isn't measuring. Similarweb tracks direct web visits to the seven platforms, not API calls, not mobile app usage, not enterprise deployments. Claude's actual usage across Anthropic's API, Amazon Bedrock, and enterprise contracts doesn't appear in these numbers. These are consumer web traffic numbers. With that caveat noted: they're still the best proxy we have for consumer mindshare, and they're telling a clear story.

Source spread

• Similarweb — Gen AI stats 2026 — builder. Primary data source; methodology is direct web-visit tracking across seven platforms, April 2026.
• First Page Sage — Top generative AI chatbots — builder. Independently reported April 2026 data with worldwide and US breakdowns.
• ALM Corp — Gemini vs ChatGPT market share — skeptic. Flags that Gemini's worldwide/US gap (27.4% vs 19.2%) traces to Android defaults in emerging markets, not US user preference.

What the data says

22pp

→ Source: Similarweb

The April 2026 worldwide breakdown across seven tracked platforms:

1. ChatGPT: 54.7% worldwide, 58.9% US
2. Gemini: 27.4% worldwide, 19.2% US
3. Claude: 8.2% worldwide, 12.5% US
4. DeepSeek: 4.1%
5. Grok: 2.8%
6. Perplexity: 1.5%
7. Microsoft Copilot (web): 1.3%

The worldwide vs US gap on Gemini is the tell. Google has more than three billion active Android devices globally, many in markets where Gemini is the default assistant. The worldwide 27.4% reflects Android defaults in Brazil, India, and Southeast Asia. The US 19.2% is what happens when users actually have to choose.

Why the share shifted

Gemini's rise is not a model quality story. In February 2025, when Gemini was at 5.6%, GPT-4o was unambiguously stronger as a product. Models have traded benchmark positions since then, but no reading of the evidence supports Gemini gaining 21 points because it suddenly got way better.

Three distribution events drove the shift.

Android defaults. Google made Gemini the default assistant on Android in May 2024. There are roughly 3.3 billion active Android devices. Press and hold the home button on most Android phones and you get Gemini. Users didn't choose this.

Google Search integration. AI Overviews in Google Search began routing hundreds of millions of Search queries through Gemini-backed responses in 2024. Those users weren't "switching to Gemini." They were getting Gemini answers on a surface they already visited daily.

Today's iOS 27 announcement. Apple just made Gemini the default AI behind Siri for iOS 27 users who configure nothing. More than 1.5 billion more devices where Gemini is the first AI you see when you do nothing. April's data doesn't include this. September's data will.

Pros & cons

What the data actually tells you:

• Default placement compounds. Each Google integration adds tens of millions of users who experience Gemini without a deliberate choice. The consumer AI attention battle is being fought at the OS and default-settings layer, not on benchmark leaderboards.
• Claude's 8.2% worldwide, and especially 12.5% in the US, is earned without any OS default anywhere. Every Claude user looked it up and chose it. That's a different kind of attention, and likely converts to retention and revenue at higher rates than traffic driven by Android home-button presses.
• DeepSeek at 4.1% globally is real traction from a lab with no default placement in Western markets.

What to hold lightly:

• Web visits are not stickiness, preference, or revenue. Google can push billions of visits to Gemini through Search defaults without a single US user actively choosing Gemini over ChatGPT in a side-by-side comparison.
• The US gap tells a different story than the worldwide number. Where choice is real and users are deliberate, ChatGPT's lead over Gemini is much larger: 58.9% vs 19.2% in the US.
• The April 2026 data is already stale. Today's WWDC announcement changes the next three months significantly.

What builders need to know

• The consumer AI market is being decided by default placements, not model quality. If your product lives in the AI attention economy, acquisition strategy has to account for the fact that most users' default AI experience is controlled by their OS vendor.
• Claude's higher US share (12.5% vs 8.2% globally) is worth noting if you're targeting US professionals. Anthropic punches above its consumer weight in the US market specifically.
• Today's iOS 27 Extensions announcement is the next variable. The September Similarweb data will be the first signal for whether a default-optional Extension converts meaningfully compared to a default-on OS integration.
• Grok at 2.8% globally, without Search integration or an Android default, is coming from X and Tesla placements. Real niche. Real model for how non-OS-owning players can hold meaningful share.
• The gap between worldwide and US numbers on any platform tells you how much of the share is earned vs installed. When in doubt, look at the US-only number first.

Further reading

• Similarweb — Gen AI stats 2026 — primary data source; methodology is direct web-visit tracking
• First Page Sage — Top generative AI chatbots by share — worldwide and US breakdowns
• ALM Corp — The shift in AI chatbot dominance — structural analysis of the worldwide vs US gaps
• Similarweb on X — Similarweb's own announcement including 12-month historical comparison