ChatGPT reached 900 million weekly active users in February 2026, according to OpenAI. Google’s Gemini app passed over 750 million monthly users, according to Alphabet. Microsoft put Copilot at 20 million paid enterprise seats. Metric class is the catch: the three figures count weekly actives, monthly actives, and paid seats, so naive stacking into one ranking misleads. Copilot and ChatGPT both run GPT-5.5, leaving Gemini as the only distinctly different model. The dimensions below cover users, pricing, models, benchmarks, and context windows across all three assistants as of mid-2026.
Key Takeaways
- ChatGPT hit 900 million weekly active users in February 2026, up 100 million from the 800 million OpenAI reported in October 2025.
- Google’s Gemini app reached over 750 million monthly active users on Alphabet’s Q4 2025 earnings, up from 650 million the prior quarter.
- Microsoft 365 Copilot reached 20 million paid enterprise seats by April 2026, a different unit from the consumer active-user counts.
- Google AI Plus at $7.99 per month is the cheapest paid entry point, undercutting ChatGPT Plus and Copilot Pro at $20 each.
- Gemini 3.1 Pro leads graduate-level science at 94.3% on GPQA Diamond, edging GPT-5.5’s 93.6%.
- GPT-5.5 leads real-world coding at 88.7% on SWE-bench Verified versus Gemini 3.1 Pro’s 80.6%.
- Microsoft Copilot and ChatGPT both run GPT-5.5, so Gemini 3.1 Pro is the only distinctly different model of the three.
Editor’s Choice
- ChatGPT counts 50 million paying subscribers alongside its 900 million weekly active users.
- Alphabet topped $400 billion in annual revenue for the first time in Q4 2025, per Alphabet.
- Only 3.33% of Microsoft’s 450 million Microsoft 365 commercial subscribers had bought a Copilot add-on at the Q2 FY26 call.
- Gemini 3.1 Pro carries a 1 million token input context window with 64,000 tokens of output.
- GPT-5.5 scored 84.9% on GDPval, a test of knowledge work across 44 occupations.
- Gemini now processes over 10 billion tokens per minute through direct API use, per Pichai.
ChatGPT Overview
ChatGPT leads on raw consumer reach, hitting 900 million weekly active users in February 2026 and 50 million paying subscribers, according to OpenAI. Late-2025 estimates put ChatGPT at around 810 million monthly active users, ahead of Gemini at that point. The product has no first-party operating system or search engine feeding it traffic, which makes the scale unusual. ChatGPT grew as a destination people choose to open, not a feature bundled into software they already run.
Under the hood, ChatGPT runs GPT-5.5, which OpenAI launched on April 23, 2026, as its most capable model. The result is a general-purpose assistant that splits the benchmark lead with Gemini depending on the task. For the wider adoption picture, our generative AI adoption data tracks how fast tools like ChatGPT moved from novelty to default.
Pros
Largest consumer reach of the three at 900 million weekly active users.- Strong agentic coding and knowledge-work benchmark scores on GPT-5.5.
- Standalone app and web access with no ecosystem lock-in required.
Cons
No native operating-system or search distribution to drive passive growth.- Premium Pro tier at $200 per month is the most expensive top tier here.
- Enterprise data-governance features lag Microsoft’s Microsoft 365 integration.
Google Gemini Overview
Gemini‘s app reached over 750 million monthly active users on Alphabet’s Q4 2025 earnings, up from 650 million the previous quarter, a sharp climb in one period. According to Google DeepMind, Gemini 3.1 Pro is a natively multimodal reasoning model and its most advanced model for complex tasks. Gemini’s growth rides Google’s existing surfaces, Search, Android, and Workspace, the opposite of ChatGPT’s standalone path. That distribution explains a chunk of the jump.
By the numbers: Pichai said Gemini’s first-party models now process over 10 billion tokens per minute via direct API use, and the Gemini app has grown to over 750 million monthly active users. Alphabet surpassed $400 billion in annual revenue for the first time the same quarter, an AI-driven milestone.
Google also rolled out a cheaper Google AI Plus plan at $7.99 per month, aimed at budget-conscious users.
Pros- Native multimodal reasoning across text, image, audio, and video.
- Cheapest paid entry point at $7.99 per month for Google AI Plus.
- Massive passive distribution through Search, Android, and Workspace.
Cons- Trails GPT-5.5 on real-world coding benchmarks like SWE-bench Verified.
- Consumer trust still recovering from earlier Gemini accuracy stumbles.
- Heavy reliance on Google’s ad ecosystem raises data-use questions for some users.
Microsoft Copilot Overview
Microsoft 365 Copilot reached 20 million paid enterprise seats by April 2026, according to Microsoft, with the figure shared by Satya Nadella on the earnings call. That figure sits against 450 million Microsoft 365 commercial subscribers, with only 3.33% having bought the Copilot add-on at the Q2 FY26 call. Copilot’s number is the odd one out: paid seats inside paying organizations, not free consumer sessions. The low penetration signals a long runway rather than a ceiling.
Microsoft prices Copilot Pro for individuals at $20 per user per month, separate from the $30-per-seat Microsoft 365 Copilot enterprise add-on.
Pros- Deep integration across Word, Excel, Outlook, and Teams for enterprise work.
- Runs GPT-5.5, the same model family powering ChatGPT.
- Enterprise data-governance and compliance controls built into Microsoft 365.
Cons- Only a small share of the Microsoft 365 commercial base has adopted it so far.
- Reported as paid seats, making direct user comparison with rivals harder.
- Less useful outside the Microsoft 365 ecosystem than standalone assistants.
Head-to-Head Comparison Table
On a side-by-side spec sheet, ChatGPT and Gemini both carry roughly 1 million token context windows, while Copilot inherits ChatGPT’s GPT-5.5 engine and its window. The table below lines up the headline numbers, with the caveat that the user-count rows measure different things.
| Dimension | ChatGPT | Gemini | Microsoft Copilot |
|---|---|---|---|
| Headline users | 900 million weekly active | 750 million monthly active | 20 million paid seats |
| Paying base | 50 million subscribers | Google AI Plus and above | Inside 450 million M365 base |
| Underlying model | GPT-5.5 | Gemini 3.1 Pro | GPT-5.5 |
| Entry price (USD/month) | 20 | 7.99 | 20 |
| Top consumer tier (USD/month) | 200 | 99.99 | 30 (enterprise seat) |
| Context window (tokens) | ~1.05 million | 1 million | ~1.05 million |
Source: OpenAI, Google DeepMind, Microsoft (2026)
User Scale: 900 Million vs 750 Million vs 20 Million
ChatGPT reached 900 million weekly active users as of February 2026, OpenAI announced, ahead of Gemini’s monthly figure and far ahead of Copilot’s paid-seat count. The gap looks enormous until you read the units. ChatGPT counts weekly actives, Gemini monthly actives, and Copilot paid seats, so a single ranking overstates the distance.
- ChatGPT: 900 million weekly active users (February 2026).
- Gemini: 750 million monthly active users (Q4 2025).
- Microsoft Copilot: 20 million paid enterprise seats (April 2026).
Copilot’s 20 million paid enterprise seats is a different metric from the active-user counts its rivals report. For a like-for-like view, ChatGPT sat at around 810 million monthly users in late 2025, closer to Gemini’s monthly figure than the weekly headline suggests.
Worth noting: The cleanest apples-to-apples line is monthly active users, where ChatGPT’s around 810 million in late 2025 sits close to Gemini’s monthly total. Gemini’s over 750 million monthly users put the two within striking distance. Copilot’s 20 million paid seats belong in a separate column, because paid seats and free actives answer different questions.
That scale tracks the same search-to-assistant shift visible in Google usage data, where assistant queries increasingly substitute for classic search.
Pricing Compared
Google undercuts both rivals on price, offering AI Plus at $7.99 per month against ChatGPT Plus and Copilot Pro at $20 each. At the premium end, the gap widens further depending on which top tier you pick.
- Google AI Plus is the cheapest paid entry at $7.99 per month.
- ChatGPT Plus and Copilot Pro both cost $20 per month.
- Google AI Ultra runs $99.99 per month.
- Microsoft 365 Copilot enterprise seats cost about $30 per user per month.
- ChatGPT Pro is the steepest tier at $200 per month.
Google recently rolled out the Google AI Plus tier at $7.99 per month. ChatGPT’s Pro tier at $200 per month is the steepest top tier of the three, aimed at heavy professional use.
Why it matters: Price tiers reveal strategy: Google’s $7.99 floor chases volume and ad-adjacent reach, while ChatGPT’s $200 Pro tier monetizes power users. Microsoft’s $30 enterprise seat sells governance and Office integration rather than raw chat. The cheapest tool is not automatically the best value once the use case is fixed.
Benchmark Performance
No single assistant wins every benchmark: Gemini 3.1 Pro leads graduate-level science at 94.3% on GPQA Diamond, while GPT-5.5 leads real-world coding, per the published model figures. The pattern matches what our AI benchmark coverage keeps showing: capability rankings split by task and shift every few months, so a clean overall winner is rarely the honest answer.
- GPQA Diamond: GPT-5.5 93.6%, Gemini 3.1 Pro 94.3% (graduate science).
- SWE-bench Verified: GPT-5.5 88.7%, Gemini 3.1 Pro 80.6% (real-world coding).
- GPT-5.5 also leads on Terminal-Bench 2.0 (82.7%) and GDPval (84.9%).
| Benchmark | What it tests | GPT-5.5 (ChatGPT, Copilot) | Gemini 3.1 Pro |
|---|---|---|---|
| GPQA Diamond | Graduate science | 93.6% | 94.3% |
| SWE-bench Verified | Real-world coding | 88.7% | 80.6% |
| SWE-bench Pro | End-to-end coding | 58.6% | not disclosed |
| GDPval | Knowledge work, 44 jobs | 84.9% | not disclosed |
| Terminal-Bench 2.0 | Command-line workflows | 82.7% | not disclosed |
| ARC-AGI-2 | Abstract reasoning | not disclosed | 77.1% |
Source: Google DeepMind Gemini 3.1 Pro model card; OpenAI GPT-5.5, 2026
The data_table above sets the scores side by side. Copilot inherits GPT-5.5’s results because it runs the same model.
GPT-5.5 posted a state-of-the-art 82.7% on Terminal-Bench 2.0 and 58.6% on SWE-bench Pro at its April 2026 launch. Gemini 3.1 Pro answered 51.4% of Humanity’s Last Exam with tools and 77.1% on ARC-AGI-2, its strongest reasoning marks.
Underlying Models and the Shared-Engine Paradox
Microsoft Copilot runs OpenAI’s GPT-5.5, the same model family behind ChatGPT (launched April 23, 2026), which means the two products share an engine. Gemini 3.1 Pro is the only one of the three built on a distinctly different model, Google DeepMind’s natively multimodal reasoning system. This reframes the contest. ChatGPT versus Copilot comes down to packaging and distribution over the same underlying intelligence.
- Microsoft Copilot runs OpenAI’s GPT-5.5, the same engine behind ChatGPT, launched April 23, 2026.
- ChatGPT and Copilot share that model, so the difference is packaging and distribution, not intelligence.
- Gemini 3.1 Pro is the only one of the three on a distinct model, Google DeepMind’s native multimodal system.
- A Copilot-versus-ChatGPT benchmark gap usually reflects configuration, not model quality.
Does Microsoft Copilot use ChatGPT? Microsoft 365 Copilot runs on OpenAI’s GPT-5.5, the same model family that powers ChatGPT. The products differ in interface, data access, and pricing rather than core model quality, so a benchmark gap between Copilot and ChatGPT usually reflects configuration and grounding, not a different brain.
Enterprise Adoption and Distribution Channels
Copilot’s enterprise engagement now matches Outlook’s weekly usage level, Satya Nadella said, with queries per user up nearly 20% quarter-over-quarter by April 2026. Large deployments are scaling: companies including Bayer, Johnson & Johnson, Mercedes, and Roche each maintain over 90,000 Copilot seats. Each assistant’s growth tracks its parent’s surface: Copilot rides Microsoft 365, Gemini rides Search and Android, ChatGPT stands alone. That predicts user capture better than any benchmark.
- ChatGPT: standalone app and web, 900 million weekly active users.
- Gemini: Search, Android, Workspace, 750 million monthly active users.
- Microsoft Copilot: Microsoft 365 and Windows, 20 million paid enterprise seats.
| Assistant | Primary distribution channel | Adoption signal |
|---|---|---|
| ChatGPT | Standalone app and web | 900 million weekly active users |
| Gemini | Search, Android, Workspace | 750 million monthly active users |
| Microsoft Copilot | Microsoft 365 and Windows | 20 million paid enterprise seats |
Source: OpenAI, Alphabet, Microsoft, 2026
The runway is wide because only 3.33% of Microsoft’s 450 million commercial subscribers had adopted Copilot at the Q2 FY26 call, leaving most of the base unconverted.
Where those seats land in the workforce is the throughline our AI job impact data follows across roles and industries.
Multimodal and Context Window Capabilities
Gemini 3.1 Pro processes up to 1 million tokens of input and 64,000 tokens of output natively across text, image, audio, and video, per Google DeepMind. GPT-5.5 carries a comparable 1,050,000-token context window with up to 128,000 output tokens. On context size, the two frontier models are close, and Copilot inherits GPT-5.5’s window. Gemini’s edge is native multimodality, which matters most for video and audio.
- Gemini 3.1 Pro processes up to 1 million input tokens and 64,000 output tokens across text, image, audio, and video.
- GPT-5.5 carries a comparable 1,050,000-token context window with up to 128,000 output tokens.
- On context size the two frontier models are close.
- Gemini’s first-party models process over 10 billion tokens per minute via direct API use.
Scale on the back end is large too: Gemini’s first-party models process over 10 billion tokens per minute through direct API use, Pichai said. Anthropic’s Claude rounds out the frontier field, and the Claude vs ChatGPT data sets a fourth model against the GPT-5.5 family.
Verdict by Use Case
The best assistant depends on the job: GPT-5.5 leads coding at 88.7% on SWE-bench Verified, Gemini leads science at 94.3% on GPQA, and Copilot wins on Microsoft 365 integration. No single tool wins outright, so the verdict is split by what you actually do with it.
- Best for coding and agentic tasks: ChatGPT, because GPT-5.5 leads real-world coding at 88.7% on SWE-bench Verified and 82.7% on Terminal-Bench 2.0.
- Best for science and research depth: Gemini, because Gemini 3.1 Pro tops graduate science at 94.3% on GPQA Diamond and adds native multimodal reasoning.
- Best for enterprise teams on Microsoft 365: Microsoft Copilot, because it runs GPT-5.5 inside Word, Excel, and Outlook with engagement now at Outlook levels.
- Best for budget-conscious users: Gemini, because Google AI Plus at $7.99 per month is the cheapest paid entry point.
Which AI assistant is best for coding?
ChatGPT, running GPT-5.5, is the strongest coder of the three on published benchmarks. It scored 88.7% on SWE-bench Verified, ahead of Gemini 3.1 Pro’s 80.6%. GPT-5.5 also reached 58.6% on the harder SWE-bench Pro. Because Copilot runs the same GPT-5.5 engine, it inherits similar coding strength inside the Microsoft ecosystem.
Which AI assistant is best for enterprise teams?
Microsoft 365 Copilot fits enterprise teams that already run Microsoft 365, with 20 million paid seats and engagement now matching Outlook’s weekly usage. Gemini suits organizations standardized on Google Workspace, while ChatGPT works for teams wanting a model-first assistant without ecosystem lock-in. The deciding factor is usually the office suite a company already pays for.
Conclusion
ChatGPT’s 900 million weekly active users, OpenAI’s headline figure, anchor a comparison of three different races rather than one leaderboard. Gemini’s over 750 million monthly users and Copilot’s 20 million paid seats round out the field on their own yardsticks. ChatGPT wins raw consumer reach and agentic coding, Gemini wins science benchmarks, multimodality, and price, and Copilot wins enterprise teams already inside Microsoft 365.
The sharpest model rivalry is Gemini against the GPT-5.5 family that powers both ChatGPT and Copilot. The practical question is no longer which assistant is best overall, but which fits the work in front of you. Distribution channels, not leaderboards, will decide the next wave.
