Google rebrands Spark as Tasks, launches free agentic AI with Gemini 4 Hydrogen
Google's Spark AI agent is now called Tasks, powered by Gemini 4 Hydrogen, and available free to users across six countries including the US.
Last verified:
Spark Transitions to Free, Multi-Region Deployment
According to The Verge AI, Google has rebranded its Spark AI agent to Tasks and removed the paywall that restricted access to paying subscribers. Originally launched in May as a feature exclusive to Google AI Pro and Ultra tiers, Tasks is now available to all users in six markets: the US, Canada, Japan, Korea, Australia, and India. The underlying system is Gemini 4 Hydrogen, which powers the agent’s ability to orchestrate multi-step workflows.
Expanded Capability Set Tailored by Region
Tasks can perform baseline functions across all supported regions: creating meal plans, managing calendar entries, discovering recurring expenses from email archives, and assigning workflows via email (gemini@google.com) or text. In the US, the agent gains additional capabilities including phone calls to businesses to compare product pricing and verify inventory levels, in-app purchase execution through Google Pay, and account sign-in powered by Google Password Manager integration.
The manual permission gate on sensitive operations—particularly purchases—reflects a design choice to maintain user consent over automated financial transactions. Users retain the ability to interrupt and approve each step of execution.
Competitive Context and Privacy Architecture
The Tasks rollout positions Google alongside other agentic offerings announced or deployed over the past year. According to The Verge, comparable systems include Claude Duet from Anthropic, OpenAI’s Dots, and Meta’s Muse. Google’s execution model differs in its reliance on hardware-isolated virtual machines, with each user’s machine instance never shared across accounts—a architecture detail that The Verge notes Google emphasizes for containment and audit.
Why This Matters
The shift from subscription-gated to free-tier access significantly broadens Google’s installed base for agentic AI, lowering the activation barrier to trial. For teams and individuals evaluating AI agents for personal task automation—expense management, travel booking, calendar optimization—Tasks now competes on reach rather than pricing alone. The regional feature parity gap (US-only phone calling and payments) signals Google’s cautious rollout of real-world transaction authority; teams outside the US should expect those capabilities to arrive on a delayed cadence. For Google’s AI strategy, the rebrand from Spark to Tasks reflects a shift toward positioning Gemini as a workflow platform rather than a conversational assistant, mirroring OpenAI’s escalation toward agent-centric product design.
Frequently Asked Questions
What is Google Tasks and how does it differ from the original Spark agent?
Tasks is Google's rebranded version of the Spark AI agent, now powered by Gemini 4 Hydrogen and available for free instead of being limited to paid subscribers. It automates multi-step workflows across email, calendar, meal planning, and business interactions.
What can Tasks do in the US that isn't available elsewhere?
US users can instruct Tasks to call businesses to compare pricing and check inventory, make purchases via Google Pay, and sign into accounts using Google Password Manager. All regions can email or text tasks to gemini@google.com.
How does Google isolate Tasks execution for privacy?
According to Google, Tasks runs in a hardware-isolated virtual machine dedicated to each user's account and never shared with other users. For sensitive actions like purchases, the agent requests explicit permission before proceeding.