Google Says Gemini 4 Could Launch Well Before Year-End as AI Race Heats Up

Google is getting ready to launch Gemini 4, its next flagship artificial intelligence model, and the release could come well before the end of the year. Koray Kavukcuoglu, the new head of Google DeepMind, told The Information's AI Agenda Live Summit this week that the model has entered the post-training phase, marking the most concrete public timeline Google has shared for its next-generation AI.

What the DeepMind chief said

Speaking publicly for the first time since taking over leadership of Google's AI research division, Kavukcuoglu said Gemini 4 is currently in early post-training, the stage where a base model is refined for more reliable real-world performance. He said Google wants to release an early version of the model as soon as possible because of what internal teams are seeing.

'Our intention is to, like, as soon as possible, to release an early post-training output because we see the results and we are excited,' Kavukcuoglu said at the summit, according to multiple reports. He added that Google plans to 'continue the fast-paced iterations' after the initial release, suggesting a strategy of shipping early and improving the model in successive updates rather than waiting for a single polished debut.

The update clears up uncertainty that has surrounded the project. In July, Google had only confirmed the start of what it called its 'most ambitious pre-training to date' for Gemini 4, with no public launch forecast. The move to post-training two months later signals that the base model is now built and entering refinement.

Why Google is in a hurry

Google has not shipped a new flagship model since the Gemini 3 series debuted in November 2025, nearly ten months ago. In that window, rivals have moved aggressively. OpenAI shipped GPT-6, and Anthropic launched its Mythos model line, both of which are widely seen as outperforming Gemini 3 and have given those companies a meaningful lead.

The gap widened further when a promised Gemini 3.5 Pro update, which Alphabet CEO Sundar Pichai had positioned in June as a bridge release, never materialized. Kavukcuoglu acknowledged that Google had 'taken a little bit of a step back' to focus on faster, cheaper Flash-class models instead, a strategy that paid off in agentic workloads but left the frontier tier exposed.

The comments come after a leadership transition at DeepMind. Demis Hassabis stepped back from day-to-day management this year to become the lab's chair and Alphabet's chief scientist, handing the Gemini 4 effort to Kavukcuoglu. Despite the lag, the new DeepMind chief told the summit he believes it is a certainty that Google will stay at the frontier of AI.

What Gemini 4 could mean for Google's products

The stakes for the launch go well beyond benchmark bragging rights. Gemini is being woven across Google's ecosystem, from Search and Android to Workspace and YouTube. A stronger flagship model would feed directly into how hundreds of millions of people use those products, from more capable AI answers in Search to smarter on-device assistants on Android phones.

Analysts are also watching the commercial strategy. Google's rivals have been landing at matching price points for top-tier models, and OpenAI and Anthropic have each emphasized cost efficiency for coding-agent workloads. If Gemini 4 delivers on its internal promise, Google could regain pricing and performance leverage it has lacked since late 2025.

Investors reacted cautiously to the news, with Alphabet shares dipping around 3.8 percent on the day of the remarks while Meta Platforms edged up, a sign that markets are weighing whether the accelerated timeline signals confidence or pressure.

Why it matters for Pakistan

For Pakistan, the Gemini 4 timeline matters in practical terms. Google's models power tools used by millions of Pakistani students, freelancers, and software developers, from Gemini in Android phones to AI features in Workspace. A more capable flagship typically flows down into free and low-cost tiers within months, putting stronger reasoning, coding, and multilingual support in the hands of users who cannot pay for premium subscriptions.

Pakistani software houses and startups building on Google's AI APIs will also be watching pricing closely. Cheaper, faster frontier models reduce the cost of running AI-powered customer support, translation, and automation services, which matters in a market where margins are thin and dollar-denominated API bills are a real constraint. If Google uses Gemini 4 to close the price gap with rivals, local developers get more room to experiment.