Google launches Gemini 3.1 Pro and retires Assistant in 2026

Google rolled out Gemini 3.1 Pro, a multimodal AI model that handles text, images, audio, video and code with a context window of up to 1 million tokens, and made it available through the Gemini app, NotebookLM and the Gemini API for Pro and Ultra users [1]. At the same time, Google announced that Google Assistant will be retired on mobile, tablets, smartwatches, headphones and cars by the end of 2026, with Gemini slated to replace it on those devices [2].

Gemini 3.1 Pro is positioned as a “hard‑core” reasoning engine for tasks that need more than a simple answer. It powers new agentic workflows, supports developers via AI Studio, Vertex AI and Android Studio, and can be accessed through the Gemini CLI. The launch is part of a broader suite that includes Gemini Enterprise for Customer Experience, which bundles pre‑built shopping and food‑ordering agents that can be deployed in days, and Agentic Vision for Gemini 3 Flash, which adds a Think‑Act‑Observe loop to let the model iteratively zoom into and manipulate images before answering [1].

The AI push coincides with a strategic shift away from Google Assistant. After years of feature removal, the company plans to retire the assistant across most consumer platforms, urging users to adopt Routines as a fallback and leaving the future of Google Home and Nest devices uncertain [2]. This move signals Google’s confidence that its Gemini models can deliver comparable, if not superior, conversational experiences.

If Gemini’s new capabilities live up to the claims, developers and enterprises will gain a unified AI backbone for both text‑heavy and visual tasks, potentially accelerating adoption of AI‑driven customer service and creative workflows. However, the abrupt phase‑out of Assistant raises questions about transition support for existing users and whether Gemini can seamlessly fill the gap across the diverse hardware ecosystem Google controls. The coming months will reveal whether the AI upgrades can offset the loss of a long‑standing voice interface.