Jack Paul 
Google says its Gemini app now has more than one billion monthly users. The company shared the number during its Made by Google event this week. It is a big milestone, but the announcement comes at a time when independent benchmarks still place Google’s top models behind rivals.
Voice and image use drive growth
Much of that growth appears to come from voice. According to Google, 63% of Gemini users talk to the assistant instead of typing. Some users rely on voice only. Busy parents are 43% more likely to use voice for everyday tasks. Image generation is also picking up. The app now produces more than 150 million images a day.
On iOS, Gemini has crossed 100 million active users. The company says macOS users are among the heaviest, sending prompts at roughly double the rate seen on other platforms. On Android, the assistant can now trigger actions inside more than 40 apps. Google calls Gemini the fastest-growing product in its history and says it is already chasing the next billion users.
Benchmarks tell a different story
The user numbers are strong. The model rankings are less flattering. On the Artificial Analysis Intelligence Index, Anthropic’s Claude Opus 5 leads with a score of 63. Google’s best-placed model, Gemini 3.6 Flash, scores 52. Models from OpenAI, xAI, and Kimi also rank above or equal to Google’s top entry.
Days before the milestone, Google shipped three new models aimed at cost and scale: Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber. The 3.6 Flash release becomes Google’s default workhorse model. Google says it runs about 17% leaner on output tokens and supports a context window above one million tokens.
Flash Cyber is a different kind of release. It is Google’s first Gemini model built to find and patch software vulnerabilities. For now, access is limited to governments and vetted partners under a restricted program.
Speed and app integrations
On speed, Google still has an edge. Artificial Analysis ranks Gemini 2.5 Flash-Lite as the fastest model it tracks, with a time-to-first-response of 0.31 seconds. But that model lags on composite intelligence benchmarks that combine coding, reasoning, and knowledge.
Google also said Gemini will connect to services like Wix, OpenTable, Ticketmaster, Pandora, Angi, and Zocdoc in the coming weeks.
Sign language model arrives
DeepMind also showed a sign-language-to-text model called SL2T. It was trained on more than 100,000 hours of data across over 50 sign languages. The model powers sign-to-text dictation in Gboard and Live Transcribe on the Pixel 11, starting with American Sign Language to English. That means Deaf users can sign to their phone in places where they would normally type.
The milestone and the benchmarks are not directly at odds. They just measure different things. Google has built an app people use, but the race on model quality is still wide open.