In May 2026, ChatGPT crossed 1 billion monthly active app users, according to market intelligence firm Sensor Tower — the fastest any app in history has reached that scale. It took ChatGPT roughly three years. Google Maps, TikTok, Instagram, and YouTube each took between five and eight.

Anthropic’s Claude app had, by the same data source, approximately 56 million monthly active users. About 5.6% of ChatGPT’s mobile audience.

That sounds like a losing position. It isn’t, necessarily — and understanding why is directly relevant to decisions about which platform you build on.

What the Numbers Actually Measure

A few caveats before reading too much into either figure:

The Sensor Tower numbers count app users, not total engagement. ChatGPT’s 1 billion figure counts people who opened the mobile app at least once in May 2026. It does not include people accessing ChatGPT through the web, API integrations, or third-party apps built on the OpenAI API. The true total-touch number is larger.

Neither figure is an audited disclosure. Sensor Tower produces market intelligence estimates. OpenAI has not confirmed 1 billion as an official metric. Anthropic has not confirmed 56 million.

Growth rate is compounding differently. ChatGPT is growing at roughly 62% year-over-year from its 1 billion base. Claude is growing at roughly 640% year-over-year from its 56 million base. These are not comparable trajectories — a high growth rate from a smaller base can persist for years before the base is large enough for absolute numbers to converge.

With those caveats noted: these are the best public estimates available, from the same source, measured the same way. The relative positioning is meaningful even if the absolute counts are imprecise.

The Session Displacement Signal

The most operationally interesting data point is not the raw user count — it’s a behavioral pattern that Sensor Tower observed:

US users who installed Claude during Q1 2026 spent approximately 5% less time on ChatGPT within one month of installing Claude.

Five percent is not catastrophic for OpenAI. At 1 billion users, even meaningful Claude adoption won’t make a dent quickly. But 5% session displacement is a real substitution signal, not just coexistence. When users who install Claude immediately reduce their ChatGPT time, that suggests they are using the products interchangeably for at least some tasks — and Claude is winning those sessions.

For builders, this is useful context: the two products are partially substitutable to end users, which means you are not necessarily capturing a larger total market by supporting both — you may be splitting a single addressable market across two backends. The question is which backend better serves your specific use case.

The Revenue Quality Divergence

Raw user count and revenue quality are not the same metric.

In April 2026, Anthropic became the top vendor by US business AI payments, according to the same Sensor Tower data. This happened while ChatGPT had 18x the user base.

This is the clearest statement of what the split actually looks like in 2026:

  • OpenAI/ChatGPT has the largest consumer AI audience in history. Their users skew consumer, casual, and varied in intent. Monetization relies on high volume at lower per-user revenue.
  • Anthropic/Claude has a smaller user base that is skewing toward business and enterprise. Their users are making consequential, recurring purchasing decisions. Monetization relies on lower volume at substantially higher per-user revenue.

Both are viable business models. They describe different markets.

What This Means If You’re Building a Consumer Product

If you are building a consumer-facing AI product — something end users encounter directly, possibly through a plugin, action, or integrated experience — the distribution math strongly favors building on or for the ChatGPT ecosystem first.

1 billion monthly users is a distribution channel. The Plugins/Actions ecosystem, even at modest conversion rates, exposes your product to an audience no other AI platform can match. If you’re building something that benefits from user volume — a discovery product, a game, a creative tool, a companion app — ChatGPT’s install base is the relevant comparison point.

Claude’s consumer base is growing fast, but 56 million is still a fraction. Consumer discovery through Claude’s app surfaces is more limited. This will likely change, but the current state favors OpenAI for consumer reach.

One important wildcard: Siri. Apple’s WWDC keynote is on Monday, June 8. The widely expected announcements include Siri 2.0 integration with both Claude and Gemini as backends, and a mechanism for users to set third-party AI services as the default for Apple Intelligence features like Writing Tools and Image Playground. If Apple ships this, Claude’s addressable surface on iOS could expand dramatically — not through the Claude app, but through the Siri and Writing Tools surface on over 1 billion active iPhone devices.

Access via Siri is different from direct app engagement. But it is a distribution mechanism that does not exist for any AI provider today, and it could shift consumer reach significantly after tomorrow’s keynote.

What This Means If You’re Building Enterprise or Developer Tools

If your product targets businesses, developers, or knowledge workers making deliberate tool choices — the revenue data suggests Claude’s ecosystem may offer better access to your target buyer.

Anthropic’s API pricing and business tier are already attracting higher-value enterprise buyers. The tools being built on Claude tend to be workflow-integrated, often involving longer context, higher accuracy requirements, or regulated environments. Claude’s usage patterns skew toward sustained, purposeful sessions rather than casual queries.

For B2B SaaS products, enterprise workflow integrations, agentic pipelines, and developer tooling: Claude is increasingly where the business buyer lives, even though it has fewer total users.

This also maps to support behavior. Claude’s customer is more likely to pay for capability, evaluate on accuracy benchmarks, and escalate support issues. ChatGPT’s customer is more likely to be price-sensitive and churn quickly. These are generalizations, but they reflect different platform cultures that affect what you can charge.

The Dual-Model Reality (and Its Costs)

Many serious enterprise teams have already settled on a practical answer: use both.

Route consumer-facing or cost-sensitive workloads to GPT-5.5-Instant. Route accuracy-critical, long-context, or enterprise-compliance-sensitive workloads to Claude. Manage the complexity in your application layer.

This works. It also has real costs:

  • Two API integrations to maintain — with different schemas, rate limits, pricing structures, and deprecation schedules
  • Two evaluation datasets — models behave differently enough that what works for one requires separate evals for the other
  • Two relationships and billing lines — administrative overhead that matters once you’re at scale
  • Context and capability drift — OpenAI and Anthropic update models on different cadences; what was true in January may not be true in June

The dual-model approach is not free. For builders who have the engineering capacity, it can offer resilience and optimization. For smaller teams, committing to one primary provider and treating the other as a fallback is often the more practical choice.

The IPO Lens

Both Anthropic and OpenAI are moving toward public markets.

Anthropic filed its confidential S-1 with the SEC on June 1, 2026, at a $965 billion valuation after its $65 billion Series H. OpenAI is expected to follow within weeks.

The IPO filings will eventually require audited financials and user disclosures — which will either confirm or revise the Sensor Tower estimates. More immediately, the IPO processes create a period of platform risk for builders:

  • Both companies will be optimizing for metrics that look good in a prospectus (ARR, MAU, enterprise penetration)
  • Pricing changes, feature launches, and API updates may accelerate as both companies compete for financial narrative
  • Lock-in matters more once a company has public shareholders to answer to — watch for enterprise contracts and commitments that may become harder to renegotiate post-IPO

If you are currently on month-to-month API agreements with either company, now is a reasonable time to review what happens to your pricing if either company goes public and repositions its enterprise tier.

The Bottom Line for Builders

Decision Lean ChatGPT/OpenAI Lean Claude/Anthropic
Consumer product, volume matters ✓ 1B users —
Enterprise buyer, LTV matters — ✓ #1 US biz payments
Developer tooling, accuracy critical — ✓
Cost-sensitive consumer workloads ✓ GPT-5.5-Instant —
Long context, complex reasoning — ✓
iOS distribution via Siri (post-WWDC) — Watch Monday

The 1 billion number is genuinely significant — no AI product has ever had this reach, this fast. But it describes one market. Claude’s 640% growth rate and enterprise revenue dominance describe a different market that is often more relevant for B2B builders.

The useful question is not “which platform is winning” — both are growing rapidly. It is “which platform’s users look like my buyers.”

Source data: Sensor Tower market intelligence estimates via Reuters, June 2026. Figures are not audited disclosures from OpenAI or Anthropic.