AI 热点

Google 升级 Gemini Task 模式,引入 Gemini 4 Hydrogen 与电商功能

SiliconANGLE更新于 11月19日 08:35

Google upgrades Gemini’s agentic Task mode with new model, e-commerce features - SiliconANGLE

AI 导读

Google 升级 Gemini Task 模式,将底层模型切换为 Gemini 4 Hydrogen,并增强多步骤任务与电商交易能力。该模型可并行执行项目步骤,部分任务处理速度较上一代入门模型快 2.8 倍,并会在网购等敏感操作前主动征求指示。Task 模式现已在美国、日本、韩国、加拿大、澳大利亚和印度免费开放,未来还计划支持支付 token 化与 UCP。

推荐理由

精选AI 评分 84

更新同时覆盖模型切换、并行执行、支付安全与免费范围,可用于评估通用智能体进入交易流程的产品路径。

主题

标签

正文

Google LLC is releasing a new version of its Gemini artificial intelligence assistant that is better at performing multi-step tasks.

Josh Woodward, the vice president responsible for the Gemini app, detailed the update in a Thursday blog post.

The AI assistant has two modes that are accessible through separate interface tabs. The first, Chat, generates quick, relatively simple responses to prompts. The second mode, which is called Task, enables Gemini to perform complex tasks such as booking appointments and comparing online product listings. Task is the focus of this week’s update.

Google is switching the feature to Gemini 4 Hydrogen, its newest large language model. It’s somewhat less capable than Gemini 4 Carbon, the search giant’s most advanced LLM, and significantly more cost-efficient. Hydrogen is also fast: Google says that it can process some tasks 2.8 times faster than its previous entry-level model.

One way Hydrogen reduces wait times is by parallelizing project steps that are usually performed one after another. According to Woodward, users of Gemini’s Task mode can give the model feedback after each step to guide its work. The model proactively asks users for instructions before performing a sensitive task such as making an online purchase.

Today, Gemini’s Task mode uses Google Wallet to process transactions. Woodward said that the search giant is working with American Express, Mastercard, and Visa to add support for a technology called payment tokenization. It replaces the user’s card number with a placeholder value called a token. That token is shared with merchants instead of the card number to reduce the risk of data leaks.

Google also plans to add support for a second technology called the Universal Commerce Protocol, or UCP. It makes it easier for AI agents such as Gemini to interact with online stores and thereby lowers the risk of errors.

Gemini’s Task mode can use data in other Google services such as Gmail to inform its decisions. For example, the agent might take the user’s past purchases into account when generating shopping suggestions. One of Gemini’s main selling points is that it can use such data without moving it off Google infrastructure, which is often not the case with competing agents.

“When you assign Gemini a task, it uses Google Cloud’s Agent Compute Engine to work in a hardware-isolated virtual machine that’s dedicated to your account and never shared with other users,” Woodward wrote. “Each virtual machine is sandboxed, so Gemini works in a separate, isolated environment that’s regularly tested by our security red teams.”

Google has made Gemini’s Task mode free in the U.S., Japan, Korea, Canada, Australia, and India as part of the update. That’s notable because Anthropic PBC and OpenAI Group PBC charge for the comparable features they offer as part of their chatbot services. Google’s move may prompt them to revise their pricing.

Image: Google

来源

SiliconANGLE