SIAgentNews.com
← All news

Google upgrades Gemini Task mode with faster model and e-commerce tools

Google is moving its complex task-handling AI to a more efficient model while adding payment security features for online shopping.

· First reported by SiliconANGLE

2.8 timesincrease in task processing speed

The short version

  • Gemini Task mode now runs on the Gemini 4 Hydrogen model, which is optimized for speed and cost-efficiency.
  • The new model can process some tasks 2.8 times faster than the previous entry-level version.
  • Google is integrating payment tokenization and the Universal Commerce Protocol to improve e-commerce security.
  • Task mode is now free in the U.S., Japan, Korea, Canada, Australia, and India.

Why it matters

This update signals a shift toward 'agentic' AI that can perform multi-step actions like shopping and booking rather than just answering questions. By offering these complex features for free in major markets, Google is putting pressure on paid competitors like OpenAI and Anthropic.

A shift toward agentic AI

Google is refining its Gemini AI assistant to better handle complex, multi-step requests through a dedicated interface. While the standard Chat mode provides quick answers to simple prompts, the new Task mode is designed for actions like comparing product listings or booking appointments.

The company is moving this specific functionality to a new model called Gemini 4 Hydrogen. While it is not as powerful as the flagship Gemini 4 Carbon model, it is specifically engineered to be more cost-efficient for the company to run. This allows Google to provide more complex capabilities without the high overhead of its most advanced systems.

As reported by SiliconANGLE, this transition aims to make the AI more useful as an agent that can actually complete work. Users can provide feedback after each individual step the AI takes to guide the process. The system is also designed to be proactive, asking for permission before performing sensitive actions like making a purchase.

Speed and parallel processing

One of the primary improvements with the Hydrogen model is its significant increase in speed. Google states that the new model can process certain tasks 2.8 times faster than the previous entry-level version it used for these functions.

This speed boost is achieved through a technique called parallelizing project steps. Instead of performing every action in a linear sequence, the AI can handle multiple parts of a task at the same time. This reduces the total wait time for the user when they assign a complex project to the assistant.

The goal of these technical optimizations is to make the AI feel more responsive during long-running tasks. By balancing the trade-off between model intelligence and processing speed, Google can offer a smoother experience for users who need results quickly. This efficiency is key for the company as it scales these agentic features to a wider audience.

E-commerce and security features

Google is also introducing specific tools to make online shopping safer and more seamless for Gemini users. The assistant can now use Google Wallet to process transactions, and the company is working with major credit card networks like Visa, Mastercard, and American Express. They are implementing a technology called payment tokenization.

Payment tokenization works by replacing a user's actual card number with a placeholder value known as a token. This token is what gets shared with merchants, which helps protect the user's financial data from being exposed during a leak. This is a standard security practice that adds a layer of protection for automated transactions.

Additionally, Google plans to support the Universal Commerce Protocol, or UCP. This technology is designed to help AI agents interact with online stores more reliably. By providing a standardized way for the AI to communicate with different websites, Google aims to lower the risk of errors during the checkout process.

Data privacy and infrastructure

The Task mode can access information from other Google services, such as Gmail, to personalize its suggestions. For instance, the AI might look at a user's past purchase history to provide better shopping recommendations. Google highlights that this data stays within its own infrastructure rather than being moved to external systems.

To ensure this data remains private, Google uses its Agent Compute Engine to run tasks. This system places the AI in a hardware-isolated virtual machine that is dedicated solely to one user's account. This means the environment is never shared with other people using the service.

Each of these virtual machines is sandboxed, which means it is isolated from the rest of the system. Google's security red teams regularly test these environments to ensure they are secure. This architecture is intended to give users confidence that their personal data is handled in a protected, private space.

Market competition and availability

Google has made the upgraded Gemini Task mode free for users in several major markets, including the U.S., Japan, Korea, Canada, Australia, and India. This is a significant move because it offers complex agentic capabilities at no extra cost to the consumer.

This pricing strategy puts direct pressure on competitors like OpenAI and Anthropic. Currently, those companies often charge a subscription fee for the more advanced features that allow their chatbots to perform complex tasks. Google's decision to offer these tools for free may force those companies to reconsider their own pricing models.

By making these tools free, Google is attempting to establish its AI as the default choice for everyday productivity. The combination of faster processing, integrated shopping tools, and a free price point creates a significant competitive advantage. The company will likely continue to refine these features as more users begin to interact with the Task mode.

What to watch

  • How competitors like OpenAI and Anthropic respond to Google's free pricing for agentic features.
  • The rollout of the Universal Commerce Protocol across more online retailers.

Try it

Check if you have the 'Task' tab available in your Gemini app to see if you can book an appointment or compare products.

Sources and links

First reported by SiliconANGLE. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

More AI news