Gemini Tasks becomes free for users in six countries
Google is opening Gemini Tasks to free users in the US, Japan, South Korea, Canada, Australia and India, after previously limiting it to paid tiers.

Key takeaways
- Gemini Tasks is free for Gemini users in six countries.
- The feature was previously limited to paid AI Pro or AI Ultra subscriptions.
- Tasks can handle background work across Workspace, email, shopping and phone calls.
- Google cites isolated browsing, purchase confirmation and clarification prompts as safeguards.
In this article
Gemini Tasks goes free
Google has made Gemini Tasks available at no cost to free Gemini users in six countries. Android Central reported the change on 18 November 2026. The feature was previously limited to paid AI Pro or AI Ultra subscriptions.
Google's Josh Woodward, vice president of product for AI, confirmed the expansion. Android Central describes the move as giving millions of people access to an AI agent. The agent can handle several steps without constant supervision.
The feature is part of Google's Gemini assistant. The rollout covers the US, Japan, South Korea, Canada, Australia and India.
What the agent can do

Gemini Tasks, previously called Spark, is designed for work that takes more than a single reply. Google says it can carry out several steps in the background. Examples include planning a trip, researching a topic, setting up a morning briefing and managing recurring tasks. The idea is to let people hand over work rather than watch each step.
The feature is also being expanded to cover shopping, phone calls, email and deeper integration with Google apps. It runs on the Gemini 4 Hydrogen model. Google describes the model as new. Google is positioning Gemini closer to a personal assistant.
In Workspace, Google says Gemini can more reliably draft or send emails. It can create and edit Google Docs and Sheets. It can also add multiple events to Google Calendar. The aim is to reduce switching between separate apps.
How to switch to Task mode
Users move from Chat to Task mode with a toggle. On mobile, the control sits above the prompt box. On the web, it appears below.
Chat mode is for ordinary conversation. Task mode is for work that continues after the user moves on.
Google also says people can bypass the app by forwarding emails or sending text messages to Gemini through Google Messages and iMessage. This gives another route for submitting work when the app is not open.
The free offer applies to the six named countries. It marks a change from the earlier model. Agentic features were previously tied to paid subscriptions.
Security controls and cautions
Android Central notes that giving an AI access to an inbox, passwords and payment methods raises security concerns. Google says Tasks uses an isolated virtual machine when browsing the web. It also cites protection against prompt injection and Google Safe Browsing checks. Prompt injection refers to attempts to mislead an AI with crafted text.
The system asks for confirmation before buying anything. It can also ask for clarification if instructions are unclear.
Google's official guidance should be reviewed before linking important accounts or payment details. The feature is still new. Users should remain cautious with sensitive tasks. The controls are meant to reduce risk, but they do not remove the need for caution.
Where the limits remain
The rollout does not make Gemini Tasks flawless. Android Central notes that some sites may prevent the agent from completing a task. They may require the user to finish the last step.
That means the assistant can still stop when a website blocks automation or demands direct action. The report describes the tool as moving closer to a personal assistant. It says the tool is not yet suitable for every task. Android Central says the prospect of an assistant reducing someone's workload is getting closer.
For readers in supported countries, the practical next step is to try the Task toggle. Users can observe where the agent succeeds or asks for help. Keeping an eye on official updates will show whether the feature becomes more reliable over time. Users should treat the agent as a helper that may need supervision.
Sources
Spotted a mistake? See our corrections policy. Read how News by OnDevice works in our editorial policy.



