At the Google I/O event in May, CEO Sundar Pichai announced that the company’s newest frontier model, Gemini 3.5 Pro, would launch in June.
“We are also excited for 3.5 Pro,” Pichai said at a pre-Google I/O media briefing. “We are using it internally. It’s showing great improvements. We are still testing and refining it, and it will roll out to everyone next month.”
That never happened.
June eventually turned into July, and then August, and as of Sept. 10, there’s still no sign of the model. The Google DeepMind website states, “3.5 Pro coming soon,” as it has for months. Meanwhile, the long-time leader of Google DeepMind, Demis Hassabis, has been shuffled into a new position.
As Google stalls on Gemini 3.5 Pro, its chief AI rivals are releasing boundary-breaking new models. OpenAI recently released GPT-6 Astra, kicking off a new generation of GPT models, and Anthropic recently released Fable 5.1. Both models have made critical advances in cyber coding capabilities. They’ve also been used to resolve famous open questions in research-level mathematics, such as the Jacobian Conjecture and the Navier-Stokes problem (the latter was resolved with quite a bit of controversy, however). Chinese AI lab Moonshot also released Kimi K3, a massive open-source model with 2.8 trillion parameters.
Meanwhile, Google has rolled out a Gemini app for Windows and lighter-weight models such as Gemini 3.6 Flash.
It’s been quite the turnaround for Google, which had raced forward in the AI arms race thanks to its massive user base and unprecedented access to training data.
So, where is Gemini 3.5 Pro?
At the Alphabet Q2 earnings call in July, Pichai said that “Gemini 3.5 Pro is currently in testing, and our team is already building the next generation of models.”
Earlier that month, Bloomberg published a report on the delayed Gemini 3.5 Pro launch, with reporters Julia Love and Davey Alba writing that “The delay has caused frustration among Google engineers, AI researchers, and managers, who are concerned the company risks losing its edge in the market to rivals Anthropic and OpenAI.”
At the time, Mashable reached out to Google with questions about the Gemini 3.5 Pro launch timeline, and the company provided the same statement it shared with Bloomberg: “We’re shipping quickly across a wide range of models while keeping them highly cost-effective for customers. We’re currently testing 3.5 Pro, an upgraded Flash model, and other models with partners, and we’re productively engaged with the U.S. government on model testing and broader frameworks.”
Mashable Light Speed
The company did not respond to Mashable’s most recent questions about the delay, and we will update this story if we receive more information.
Since the Bloomberg report, Google DeepMind has undergone a seismic leadership shakeup. In August, Pichai announced that Demis Hassabis, Google DeepMind’s co-founder, will henceforth serve as the chairman of Google DeepMind and chief scientist of Alphabet, with AI researcher and longtime Googler Koray Kavukcuoglu leading the lab as the new senior vice president. The announcement also revealed that legendary AI researcher Jeff Dean would be leaving Google to start a new venture, along with several other researchers.
Crucially, Kavukcuoglu is based in Google’s HQ in Mountain View, California, whereas Hassabis was based in London. An in-depth Fortune report said Google employees were “shocked” by the news, which raised concerns that Pichai was consolidating power in Mountain View.
So, is Hassabis’s new role a promotion or a demotion? That’s the billion-dollar question analysts have been asking ever since, though Google characterized the move as a promotion.
Fortune’s report on the “difficult year” at Google DeepMind also cited “delayed models, high-profile exits of talent, and a very public fight” among employees over military defense contracts.
Bloomberg’s July report suggested there were two other reasons driving the delay of Gemini 3.5 Pro. The first is bureaucratic. Given Google’s organizational size and the number of products integrated with Gemini, delays are inevitable compared with leaner AI startups. Second, Bloomberg found that Google leaders were worried that Gemini 3.5 Pro may not be competitive with their rivals’ recent releases.
Google continues to fall on AI leaderboards
Without a new frontier model, Google has tumbled down AI leaderboards. On the Arena leaderboard, Google is in 14th place as of this writing, behind models from Chinese labs such as DeepSeek and Moonshot. In Arena’s coding leaderboard, the area where Anthropic and OpenAI have reported such rapid advancements, Google is currently ranked 25th.
Of course, the AI industry can and does change rapidly, and a new, highly advanced model could reverse Google’s momentum overnight. The company still has some key advantages over other AI companies. Google not only has unprecedented access to the world’s data, but it can also put its AI tools directly into the hands of billions of Android and Google Search users worldwide.
And even though Gemini 3.5 Pro is delayed, the company is already talking about Gemini 4.
“We are now training Gemini 4, and we’re being very ambitious with it,” Pichai said at the Q2 earnings call. “I am very excited by the progress I’m seeing internally on Gemini 4, and I’m confident that people will be pleased when we are putting it outside.”
In response to a question about whether Google could still compete at the frontier level, Pichai sounded very optimistic.
“We are really focused, putting a lot of effort into Gemini 4. It’s a very ambitious effort. We want to compete at the frontier level of where the frontier will be when Gemini 4 comes out, and so we are applying a lot of our compute and effort in that direction…And so, picking up pace and releasing models almost at a monthly cadence is part of our roadmap as we are building Gemini 4 as well.”
But the longer Google waits to release Gemini 3.5 Pro, the higher the performance bar it will have to clear to maintain equal footing with its rivals.