Google's August AI Blitz: Gemini 3.7 Flash, Pixel 11, and a Billion-User Chatbot

Key points

• Google released Gemini 3.7 Flash in August 2026, a coding-focused AI model priced at half the cost per million tokens of the version it replaced three weeks earlier.
• The Gemini app crossed 1 billion monthly users, which Google calls the fastest-growing product in its history.
• Pixel 11, Pixel 11 Pro, Pixel 11 Pro XL and Pixel 11 Pro Fold launched with the new Tensor G6 chip running Gemini Nano on-device.
• Gemini now generates more than 150 million images per day, with small businesses among the heaviest users.
• Google open-sourced WeatherNext 2, a cyclone-forecasting model detailed in a Nature paper.

August was a big month for Google's AI push, and if you only clock one thing, clock this: the Gemini app now has more than a billion people using it every month. That makes it the fastest-growing product Google has ever shipped.

I cover games and poker for a living, so I look at numbers like that the way I look at a prize pool. A billion monthly players is bigger than any live service game on Earth. It changes what "mainstream AI" actually means.

Here is what Google, in its own monthly roundup, said it shipped.

What actually launched in August?

A cheaper coding model, a new phone line, a smarter transcription tool, and a video generator. Google released Gemini 3.7 Flash, its workhorse model for developers building software and AI agents (software that carries out multi-step tasks on your behalf). It landed just three weeks after version 3.6, at half the price per million tokens, the small chunks of text the model charges for.

The headline consumer launch was the Pixel 11 series at Made by Google 2026: Pixel 11, Pixel 11 Pro, Pixel 11 Pro XL and Pixel 11 Pro Fold. All four run the new Tensor G6 chip, which is designed to run Gemini Nano, a small version of Google's AI model, directly on the phone rather than in the cloud.

There is also Gemini 3.5 Transcribe for real-time speech-to-text, and Gemini Omni 1.1 Flash for video generation with 4K upscaling and scene extension.

Should ordinary users care?

Yes, because Gemini is quietly becoming the default assistant on Android. Google says 63% of Gemini users now talk to it by voice, and busy parents are 43% more likely to use it for everyday tasks like planning meals or juggling calendars.

Gemini Live added Personal Intelligence, Daily Brief, Spark, and hands-free inbox management. Translation: you can ask it to triage your email while you make breakfast.

For students, Google is giving eligible college students a free year of its Google AI plan, plus SAT, ACT, GRE and LSAT practice quizzes generated inside Search.

What does this do to the fun?

Here is where I get twitchy. A billion people talking to the same assistant means a billion people getting nudged toward the same answers, the same image style, the same holiday itinerary. Gemini now spits out 150 million images a day. Small businesses are using it to make marketing materials end to end.

That is genuinely useful if you run a corner shop. It is also a monoculture in the making. The joy of the early internet was weird stuff made by weird people. When the default creative tool has one house style, everything starts to rhyme.

What about the science bits?

Two stand out. Google's WeatherNext 2, published in a Nature paper, predicts cyclone tracks and intensity with what Google's researchers call a decade of forecasting progress in one model. Google is open-sourcing it.

Operation Blue Skies, a partnership with the UK Government, uses AI forecasts to help pilots avoid making contrails, the white streaks planes leave behind, which trap heat in the atmosphere. It is now expanding across the North Atlantic.

That is the underdog-versus-the-machine angle I actually like: the machine helping the atmosphere win.