Google’s Tasks is the only agent in this comparison that is free, keeps working while you do something else and can phone a business, though only in the US and only for part of the job. Google made Tasks, the agent it previously called Spark, free in six countries on November 17, 2026. The rivals differ on price, supervision and reach.
What Google changed on November 17
Google’s announcement describes Tasks as a Gemini mode for multi-step errands. You switch the home-screen toggle from Chat to Task, text Gemini, or forward an email. The task runs in a hardware-isolated virtual machine on Google Cloud, which is why it can continue after you close the app. The rollout covers the US, Japan, Korea, Canada, Australia and India and uses the Gemini 4 Hydrogen model, which we covered when it launched.
What is not global matters most. Calling businesses, texting and emailing Gemini, and Google Pay checkout start in the US. Calling is unavailable in Indiana, Louisiana, Minnesota, Montana and Nebraska, texting needs a US number with an eligible carrier, and American Express support is described as coming soon. Outside the US you get the core agent without those extras.
The real split: supervised or unattended
Price is the first difference. Supervision is the second. Claude in Chrome, according to a guide refreshed on November 13, 2026, cannot run on a schedule or unattended, and each session needs you present. OpenAI has retired ChatGPT agent mode in favor of ChatGPT Work, which targets longer workplace projects and asks for authorization before sensitive actions. Google’s Tasks and Microsoft’s Copilot Tasks are built the other way: you hand over a goal and come back to a result, with approval prompts for money.
Unattended execution is worth having only when the agent is reliable. Google’s own post says tasks can fail or miss the mark and some sites make Gemini hand the last click back to you. A free agent that fails half the time on your bank’s portal is no cheaper than doing the job yourself, so run your first tasks where a failure costs nothing.
Who each option fits
Households already on Prime get Alexa+ at no extra cost in the US, and it reaches Echo devices where Gemini does not. People who live in Google’s apps, calendar and Pay get the most from Tasks. ChatGPT users get Work, which fits documents and workplace projects better than household errands. Comet suits people who want an agent in a browser and no subscription. Claude in Chrome suits cautious users with a dedicated browser profile.
Where this is heading
Two paths are visible. One puts the agent inside a platform you already use, such as Google, Amazon or Microsoft, and bundles it free or into a subscription. The other puts it in a browser, where it works on any site but needs you nearby. Google’s move to free pressures the paid middle: an agent that costs money has to beat a free one on reliability, and nobody here has published figures that settle that.