谷歌在 I/O 2025 大会上发布了一系列 Gemini 更新和功能，旨在将其定位为用户全方位的 AI 助手。新功能包括 Sora 竞争对手 Flow、AI 研究助手升级以及免费的 Gemini Live。不过，许多新功能都包含在付费订阅计划中，如 Google AI Pro 和 AI Ultra。此次更新展示了 AI 日益增强的代理能力，即 AI 工具可以代表用户执行任务。Flow 是一款 AI 电影制作工具，Gemini 现在可以直接在 Chrome 浏览器中使用，Gemini Live 现已免费向 iOS 用户开放。此外，Gemini Deep Research 和 Canvas 也得到升级。部分新功能需要付费订阅才能使用。

🎬 **Flow：谷歌的 AI 视频生成器**：Flow 是谷歌推出的 Sora 竞品，结合了 Veo 3（具有原生音频的视频生成模型）和 Imagen 4（改进了文本和排版生成）两个版本。Flow 具有编辑功能，如组合场景、从特定相机角度和镜头“拍摄”以及修剪或扩展场景，AI Pro 用户每月可生成 100 个视频，AI Ultra 用户有更高的限制。

🌐 **Gemini 集成至 Chrome 浏览器**：Chrome 用户现在可以直接在浏览器中使用 Gemini，方便用户在阅读网页时澄清复杂信息或总结信息。此功能适用于 AI Pro 和 AI Ultra 订阅者。

📱 **Gemini Live 免费向 iOS 用户开放**：Gemini Live 使用实时摄像头和屏幕共享，之前仅包含在每月 20 美元的 Gemini Advanced 计划中，现在所有 Gemini 应用用户均可免费使用。用户可以将摄像头对准需要帮助的事物，或询问屏幕上的内容，Gemini 将以语音模式回答。

🔎 **Gemini Deep Research 和 Canvas 升级**：Deep Research 允许用户上传自己的文档，以获得更详细的研究报告。Canvas 也得到升级，可以将信息转换为网页、信息图表、测验和音频概述。这些功能由 Gemini 2.5 模型提供支持，适用于 AI Pro 和 AI Ultra 订阅者。

💼 **Gemini 赋能 Google Workspace**：Google Workspace（Gmail、Docs、Slides、Vids、Meet 等）迎来更新，包括可定制的 AI 头像、Gmail 中的个性化电子邮件回复、Gmail 中对电子邮件进行排序和组织、Google Meet 中的实时语音翻译、Google Docs 的新 AI 写作工具以及 Google Vids 的新功能。

At Google I/O 2025, the company revealed a bunch of new Gemini updates and features that position the app as your AI assistant for practically everything.

Today, Google announced a new Sora competitor called Flow, updates to your AI-powered researcher, and free Gemini Live. Free is the operative word here, since many of the other features are bundled into the paid subscription plans — Google AI Pro (formerly AI Premium) for $20 a month and a new plan called AI Ultra for a whopping $250 a month.

The features showcased today demonstrate AI's increasingly agentic capabilities, in other words, AI tools that can perform tasks on your behalf. "This is our ultimate goal for the Gemini app," said Google Labs and Gemini lead Josh Woodward in a pre-event briefing, "an AI that's personal, proactive and powerful."

Here are all the major Gemini AI announcements from today's event.

Introducing Flow, Google's Sora competitor

Google Gemini now has its own version of OpenAI's AI video generator Sora. Today, Google announced an AI filmmaking tool called Flow. This tool combines two versions of Google's media models both unveiled today: Veo 3, Google's video generating model that has native audio, and Imagen 4 which Google says has improved text and typography generation. Flow comes with editing capabilities like combining scenes, "shooting" from specific camera angles and "lenses," and trimming or expanding scenes.

Flow is available today for AI Pro, which gets you 100 generations per month, and AI Ultra with higher limits.

To show off its capabilities, Google tapped filmmaker Junie Lau to create a short film using Flow, which you can watch on YouTube.

Gemini comes to Chrome

Starting tomorrow, Chrome users will get the Gemini experience directly in their browser for Windows and macOS. This version "allows you to easily ask Gemini to clarify complex information on any webpage you're reading or summarize information," according to the announcement. Gemini in Chrome is available for, you guessed it, AI Pro and AI Ultra subscribers.

Google was found guilty of being a monopoly in an antitrust case brought forth by the U.S. Department of Justice. The case is in its remedial phase, which means Chrome might be divested from Google. What happens next is unclear, so keep an eye out for Chrome's future.

Gemini Live now free for iOS users too

Starting today Gemini Live will be free to iOS users. Recently, Gemini Live, which uses real-time camera and screen sharing, was was made available for free to Android users after previously bundling it with its paid Gemini Advanced plan for $20 a month. Now this feature is free for all users with the Gemini app.

Google touts Gemini Live as an easy, hands-free way to troubleshoot problems since you can point your camera at something you need help with or ask it questions about something on your screen and it will talk to you in voice mode.

Gemini Deep Research and Canvas get some upgrades

Google announced Deep Research for Gemini last December, and now you can upload your own documents to the feature that's described as your very own research assistant. Deep Research is basically Gemini on steroids. It crawls the corners of the web to provide a detailed report on your query. Today, Google shared that you can now upload your own personal files to the research report.

On a related note, you can add this research to Gemini Canvas, which has been upgraded to be able to transform information into web pages, infographics, quizzes, and audio overviews. Gemini Canvas will now be available in 45 languages, Google shared.

All of these features are now powered by Gemini 2.5 models. Gemini Deep Research and Canvas are available to AI Pro and AI Ultra subscribers starting today.

Access to certain new Gemini features come with a cost

Aside from Gemini Live for Android and iOS users, the other Gemini features are only available through paid subscriptions. The Google AI Pro plan costs $20 a month and will get you access to Gemini Deep Research, Canvas, and a version of Flow that uses Veo 2.

For access to Flow with Veo 3, along early access to other AI features, a YouTube Premium subscription, and 30 TB of storage there's the new Google AI Ultra plan, which costs $250 a month.

New Gemini-powered Workspace features

While this isn't technically a new feature in the Gemini app, it is an updated powered by Gemini. We have a whole article breaking down all the changes coming to Google Workspaces (Gmail, Docs, Slides, Vids, Meet, et al), but to summarize:

Customizable AI avatars that can present Google Slides presentations

Personalized email replies in Gmail

The ability to sort and organize emails in Gmail

Real-time speech translation in Google Meet

New AI writing tool for Google Docs

New features in Google Vids for video creation and editing