Tasks in Gemini Is Now Free: Which Countries Get It and What You Still Need
Google made Tasks in Gemini free on November 17, 2026 in six countries: the US, Canada, Australia, India, Japan and South Korea. Everywhere else it still needs Google AI Pro or Ultra, and it stays blocked in the UK, the EEA, Switzerland and Nigeria. You must be 18 or over on a personal Google account with history on.
Google stopped charging for Tasks in Gemini. On November 17, 2026 it made its background AI agent free of charge in six countries -- the US, Japan, South Korea, Canada, Australia and India -- running on the Gemini 4 Hydrogen model. Outside those six it still needs Google AI Pro or Ultra, and it is still blocked outright in four places. The feature itself is the one Google called Gemini Spark until mid-October 2026. Google's own help article now carries the question readers are asking -- "Where did Spark go? I can't find it." -- and answers it: "we renamed it Task to make it easier to understand. You can find it in the Chat and Task toggle next to the prompt text box. Your existing tasks and schedules are still there." For three weeks in October it also demanded that you opt in to Personal Intelligence first. Google has taken that requirement back off its list, so the gates today are the ones they were in September.
Key Takeaways
- The name changed, nothing you pay changed. Google's help article is now titled "Manage your tasks & workflows in Gemini Apps" and the word Spark survives on it exactly twice, both inside a FAQ explaining the rename. Read live October 16, 2026; the archived October 9 copy was still titled "Use Gemini Spark" and used the word 101 times.
- You do NOT have to opt in to Personal Intelligence, and Google quietly reversed itself on this. The requirements list gained "Opt in to Personal Intelligence" in mid-October 2026 and has since lost it again. Read live November 9, 2026, Google's list under the words "To assign Gemini tasks, you must:" has exactly four items -- 18 or over, personal Google Account, Google AI Pro or Ultra, Keep Activity on -- and Personal Intelligence is not one of them. The Internet Archive's October 25, 2026 capture still carries the line, so the removal falls between October 25 and November 9. Google announced neither the addition nor the removal.
- Gemini asks permission in the task thread, and one of the four answers is stickier than it looks. Per Google's permissions article, "Allow for this chat" approves an action for that thread "without changing your Connected Apps settings", while "Always allow" approves it "for the current thread and all future task threads" and automatically rewrites that action to "Always allow" in your settings. See the controls section below.
- If it uses the cloud browser, it keeps your logins. Google's words: "browser data, like cookies that contain your website authentication info (your sign-in details), is saved for future sessions for your convenience." There is a one-click wipe, and it is not on by default.
- FREE IN SIX COUNTRIES SINCE NOVEMBER 17, 2026, AND THIS IS THE HEADLINE. Google's tasks help article, read live November 18, 2026, now reads: "Available without a Google AI subscription in Australia, Canada, India, Japan, South Korea, and the US. Available with a Google AI Pro or Ultra subscription everywhere else Gemini Apps are supported, except in the European Economic Area, Nigeria, Switzerland, and the United Kingdom." The plan requirement that used to sit in the list above it is gone.
- Three more Google surfaces say the same thing, so this is not one page's wording. The announcement (
article:published_time2026-11-17): "rolling out free of charge for hundreds of millions of people across the U.S., Japan, Korea, Canada, Australia, and India, using our new Gemini 4 Hydrogen model." The release note dated 2026.11.17 repeats it. And Google's feature-availability table now ticksTasks in Geminiin its "Without a Google AI plan" column, where that cell was blank on November 16. - What the paid plans buy now is quantity, not access. Google One's plan page used to put "Access to Tasks in Gemini" on the Pro card. Read November 18, 2026, its comparison row reads "Tasks in Gemini: More / Expanded / Higher / Highest" across Plus, Pro and the two Ultra tiers -- the grading language Google uses for things everyone has some of. Its new footnote 2 reads "Availability of Tasks in Gemini varies by country." Note that this matrix has no free column at all, so it does not answer the free question either way; the four surfaces above do.
- Which model you get still depends on what you pay. Google's models table, read November 18, 2026, gives the "No Google AI plan" row
Hydrogenin task mode -- on November 15 that cell readN/A. Pro adds Argon, Ultra adds Carbon. So a free account gets a working agent on Google's cheapest current model, not the frontier one. - Still four places blocked outright: the European Economic Area, the United Kingdom, Switzerland and Nigeria. That sentence is unchanged word for word, so a UK or EU reader on Ultra still does not have it.
- Where to click has moved. It was "Switch to Spark" in the sidebar. It is now a Chat and Task toggle below the text box, and the settings are under Settings, then Task settings. The off switch is labelled "Use task mode to hand off work".
- It can now hand you finished files. New since October 9: Gemini can produce PDFs, Word documents, spreadsheets, slide decks, images and more as the output of a task, and edit Markdown and text files in Canvas. Google publishes the full format list.
- Skills are now summoned with
@, not/. The help article's invocation line changed from/to@, which closes a change Google had been signalling as "soon" since skills replaced Gems. - Google announced none of this.
gemini.google/release-notes/still stops at 2026.10.13 with zero mentions of "Tasks in Gemini"; the Gemini Apps Privacy Hub carries "Last updated: October 13, 2026" and the line "Learn more about Tasks in Gemini (formerly known as Spark)". The rename landed as an in-place edit to the documentation. - One Google page still says Spark. The Gemini Drops carousel still reads "Gemini Spark is available everywhere." The subscriptions page and Google One page have both already swapped the name.
- Plus and Pro also get a lighter feature: Daily Brief, a once-a-day personalized morning digest. Any paid tier including $4.99 Plus qualifies, per Google's subscriptions page, but Google restricts it to people 18 or over in the United States, per its help article.
- The honest verdict: if you are in one of the six countries, this is now worth ten minutes of your time because it costs nothing, and the only real question is whether you trust an agent with the apps you connect to it. If you are in the UK or the EU, it is still not a decision you can make at all. Personal Intelligence remains a separate choice rather than the price of entry, so handing Gemini your Gmail and Photos is optional.
What Changed Besides the Name
Four things, and one of them is a new requirement. Everything else about the feature -- the plans, the age gate, the four blocked countries, the apps it connects to -- is identical to what it was before Google swapped the name. This table is the whole diff, taken from Google's help article read live on October 16, 2026 against the archived copy of October 9, 2026.
| What moved | Before (October 9, 2026) | Now (October 16, 2026) |
|---|---|---|
| The name | "Gemini Spark", used 101 times on the help page | "Tasks in Gemini", or just "task mode". Spark survives twice, both explaining the rename |
| Getting in | "Opt in to Personal Intelligence" was not a requirement | It was added in that week -- and removed again between October 25 and November 9, 2026, so the list is back to four items. See Corrections and Updates |
| Where you click | "Switch to Spark" in the sidebar | A Chat and Task toggle below the text box |
| Settings | Settings and help, then "Gemini Spark Settings" | Settings, then "Task settings". The off switch reads "Use task mode to hand off work" |
| Summoning a skill | Type / | Type @ |
| What you get back | Progress, files it read, a thread | The same, plus files it makes for you: PDFs, .docx, spreadsheets, .pptx decks, images, archives |
The file-creation addition is the one worth knowing about if you already use this. Google now publishes a format list, and it is long: .pdf, .docx, .csv, .xlsx, .pptx, .md, .txt, .json, images, audio, video and compressed archives, plus Google Docs, Sheets and Slides. Markdown and plain text open straight into Canvas, where you can edit them; everything else arrives as a link you can open and download. So a research task can now end with a finished slide deck instead of a wall of chat.
Nothing you were told before about who gets this has changed. Google did not announce the rename: gemini.google/release-notes/ has no entry for it, and the only dated evidence is the Gemini Apps Privacy Hub carrying "Last updated: October 13, 2026" alongside a new line reading "Learn more about Tasks in Gemini (formerly known as Spark)". The help page was still titled "Use Gemini Spark" in the Internet Archive's October 9, 2026 capture, so the change landed in the week between.
What Is Tasks in Gemini, in Plain Terms?
Tasks in Gemini is a 24/7 personal AI agent that does real work on your behalf instead of only answering questions. Google describes it as a shift from an assistant that "can answer your questions into an active partner that does real work on your behalf and under your direction." Because it runs in Google's cloud, it keeps working even after you close your laptop or lock your phone -- the task continues without you babysitting it. It connects to the apps you already use, like Gmail, Docs, and Slides, and to outside services through app connections.
The everyday version: you give Tasks in Gemini a goal or a standing instruction, and it grinds through the steps in the background. Google's own examples include setting a recurring task to parse your monthly credit card statement and flag new or hidden subscription fees, organizing your inbox, and teaching it new routines you repeat often. It is the same "agent" idea now shipping across the industry -- ChatGPT agent mode and Claude in Chrome do comparable things -- packaged as a native part of the Gemini app.
Do You Get Tasks in Gemini? Availability by Plan
The short answer: if you are in the US, Canada, Australia, India, Japan or South Korea you get it whatever you pay, and outside those six you need Google AI Pro or Ultra. Google changed this on November 17, 2026 and the current terms live on its tasks help page rather than only in the announcement. Here is who gets what, re-read live on November 18, 2026:
| Plan | Tasks in Gemini | Chrome auto browse | Daily Brief (the lighter morning digest) |
|---|---|---|---|
| Gemini (free) | Yes in the US, Canada, Australia, India, Japan and South Korea; no elsewhere | No | No |
| Google AI Plus | Yes, and more of it | No | US only, 18+ |
| Google AI Pro | Yes, expanded | US only for now | US only, 18+ |
| Google AI Ultra | Yes, highest | US only for now | US only, 18+ |
| Business / Workspace | No -- work and school accounts are excluded for now | No | No -- personal accounts only |
The requirements list lost the plan line on November 17, 2026, and it is worth seeing what is left. Read live on November 18, 2026, the block is now headed "What you need" rather than "Requirements", and under "To assign Gemini tasks, you must:" there are three items: "Be 18 or over", "Sign in to the Gemini app with a personal Google Account" (work and school accounts are still excluded), and "Have History on". The fourth item, "Have a Google AI Pro or Ultra subscription", is gone from that list and now appears below it as a country condition instead.
One wrinkle in that third item, and it is Google disagreeing with itself about a setting's name. This site calls that setting Keep Activity, because that is what Google called it, and the tasks article now calls it History. On the same day, Google's Create and manage skills article still says Keep Activity twice, read live November 18, 2026. Same switch, two names, depending on which Google help page you open. In the app, look for whichever of the two your settings screen shows; there is only one control.
And four regions are still excluded outright, at any price: Google says Tasks in Gemini is available with a subscription "everywhere else Gemini Apps are supported, except in the European Economic Area, Nigeria, Switzerland, and the United Kingdom." July's Gemini Drop carried exactly that list in its footnote. So a UK or EU reader on Ultra still does not have it.
Personal Intelligence was a requirement for about three weeks and is not one now. Personal Intelligence is the opt-in setting that lets Gemini draw on your Gmail, Contacts, Photos, Search history and YouTube history. Google added the line "Opt in to Personal Intelligence. Learn more about Personal Intelligence." to the tasks requirements list in mid-October 2026, and has since deleted it. The Internet Archive's October 25, 2026 capture of the help article still has it; the live page on November 9, 2026 does not, and a word-for-word comparison of the two shows those eleven words are the only difference in the whole block. Google published nothing either time: gemini.google/release-notes/ has no entry past 2026.10.13.
What that means for you, in order. If you turned Personal Intelligence on because you were told you had to, you can turn it back off and still assign tasks. If you had deliberately left it off and skipped this feature, it is open to you again. And if you want the benefit without the whole bundle, the help article is clear that Gemini only "can use" Personal Intelligence for a task, listing it beside Connected Apps, skills, chats and location as one source among several -- not a switch you have to throw first.
The rename reached Google's sales pages too, and the Ultra sentence that used to confuse people is now half gone. Google's subscriptions page, read October 16, 2026, sells Ultra on "first access to advanced features like Tasks in Gemini (in select countries)" -- the same sentence it has carried since July with the new name swapped in, and with "Deep Think" no longer in it anywhere on the page. Google One's plan page used to be clearer, and has since changed its mind about what it is selling: on October 16 "Access to Tasks in Gemini" sat on the Pro card with no equivalent line on AI Plus, and on November 18, 2026 every paid card carries a graded version instead -- "More access" on Plus, "Expanded access" on Pro, "Higher access" on Ultra. Both pages now use the new name; the word Spark appears on neither. Google's August 13, 2026 Gemini 3.7 Flash post describes the feature, under its old name, as "available to Google AI Pro and Ultra subscribers in over 160 countries."
Where It Runs, and Google's Own Disagreement About Windows
Tasks in Gemini definitely runs in three places: the Gemini mobile app, the Gemini app on Mac, and the web app at gemini.google.com. Whether it runs in the new Gemini app for Windows is genuinely unclear, because Google's own surfaces disagree, and we would rather tell you that than pick one.
| Google surface | Read on | What it says about Tasks in Gemini on Windows |
|---|---|---|
| Tasks help article | Re-read Oct 16, 2026, renamed but otherwise unchanged | Tasks are "only available in the Gemini mobile app, the Gemini app on Mac, and the Gemini web app at gemini.google.com". The word Windows does not appear on the page. |
| Windows app help article | Re-read Oct 16, 2026 | Lists what the Windows app does -- answers, drafting, summarizing, brainstorming, coding help, image analysis. Neither "Spark" nor "task" appears anywhere on it. |
| Gemini release note 2026.09.10 | Sept 11, 2026 | Describes the Windows app's features, including Gmail and Drive connections and Nano Banana images. Does not mention Spark or tasks. |
| Windows launch post | Sept 11, 2026 | "Hand off multi-step tasks to Gemini Spark, your 24/7 personal AI agent." Written before the rename, and still carrying the old name. |
| Gemini Drops | Re-read Oct 16, 2026, still on the old name | A carousel card still reading "Gemini Spark is available everywhere." / "Get things done 24/7, even if you close your laptop, with Gemini Spark, now available globally." Names no surface and no excluded region. |
So it is three documentation surfaces to two marketing ones, and the marketing line is the looser one: "available globally" is not what the help page says, which on an October 16, 2026 re-read still excludes the European Economic Area, Nigeria, Switzerland and the United Kingdom by name. Marketing pages say a feature exists; help pages say who gets it. Follow the help page. What to do about it: if you are on Pro or Ultra and you install the Windows app, look below the prompt box for the Chat and Task toggle. If it is there, the launch post is right and the help page is lagging. If it is not, use the web app instead, which definitely works from the same PC. Either way the plan, age and country gates above are identical.
For current Ultra and Pro pricing, check Google's subscriptions page directly, since Google has been adjusting its AI plan prices through 2026.
Is Tasks in Gemini Free?
In six countries, yes, as of November 17, 2026. If you are in the US, Canada, Australia, India, Japan or South Korea, 18 or over, and signed in to the Gemini app with a personal Google Account, you can assign Gemini a task without paying anything. That is Google's own wording on its tasks help article, read live November 18, 2026: "Available without a Google AI subscription in Australia, Canada, India, Japan, South Korea, and the US."
Everywhere else, no, and it has not got cheaper there. Outside those six, Tasks in Gemini still needs Google AI Pro or Ultra, and in the European Economic Area, the UK, Switzerland and Nigeria you cannot buy it at all.
This is the fastest a Google consumer AI feature has gone from top-tier-only to free in 2026, and the trajectory is worth seeing in one line:
| When | Who could use it |
|---|---|
| Until July 30, 2026 | Google AI Ultra, and outside the US only Ultra |
| July 30, 2026 | Google AI Pro too, in "over 160 additional countries" |
| November 17, 2026 | Free of charge in six countries; Pro or Ultra everywhere else it runs |
What you give up by not paying is quantity and model, not the feature. Google's models table gives a free account Gemini 4 Hydrogen in task mode; Pro adds Argon and Ultra adds Carbon. Google's plan page grades the rest as More, Expanded, Higher and Highest access without publishing a single number behind any of those four words, so you cannot work out in advance how many tasks a free account gets. The help article does publish one hard cap that applies to everybody: 15 tasks running at a time, and a schedule will not fire while 15 are already going.
The question this page had open is now closed, and the answer was better than expected. On November 9 this page asked whether the agent would ever reach Google AI Plus at $4.99 and advised readers not to plan around it. It skipped Plus entirely and went to $0.
What Can Tasks in Gemini Do for a Normal Person?
Tasks in Gemini is best at the tedious middle of a task -- the clicking, sorting, and following-up you would rather not do yourself. It shines on standing, repeatable jobs rather than one-off questions. Based on Google's own announcement, here is what it targets:
| What you want | What Tasks in Gemini does |
|---|---|
| Stop losing money to forgotten subscriptions | Reads your monthly card statement and flags new or hidden recurring charges |
| Start the day organized | Works with your email and calendar to surface and prioritize what needs attention |
| Handle repeat chores | Runs recurring tasks and triggers you set once, in the background |
| Get things done across apps | Connects to Gmail, Docs, and outside services to carry a task across several steps |
| Keep control | Pauses to ask before high-stakes actions like paying or sending a message |
Google also says more abilities are coming, including texting and emailing Tasks in Gemini directly and creating custom sub-agents for specific jobs. The "operate your local browser" item on that list has now shipped: see the next section. What Tasks in Gemini is not, yet, is a hands-off robot that runs your life -- like every 2026 agent, it is slower than doing simple things yourself and needs supervision on anything that matters.
It Can Drive Your Own Chrome Browser
As of July 30, 2026, Tasks in Gemini can use your actual desktop Chrome, signed in as you. Google calls it Chrome auto browse, and with your permission it can use your logged-in accounts and the passwords saved in Chrome's Password Manager to get through login walls -- scheduling apartment viewings, researching flights and starting a booking. Google says it is "initially rolling out in the U.S. with plans to expand to additional regions in the future," and the help article, re-read October 16, 2026, puts it plainly: "Currently, Chrome auto browse is only available on desktop." On a phone, Gemini uses a remote browser in Google's cloud instead.
This is the single biggest change to what Tasks in Gemini can reach, and the single biggest thing to be careful about, so it has its own page: Chrome auto browse explained covers the local-versus-remote browser distinction, the three approval gates, and how to switch it off.
Which Apps Can It Actually Connect To?
An agent is only as useful as the apps it can reach, and its list grew in Google's June 30 update. It now reaches beyond Google's own products into a handful of named third-party services, with an open door for the rest. Here is the current picture:
| Connection | What it unlocks |
|---|---|
| Gmail | Reading, sorting, and drafting from your inbox (it asks before sending) |
| Google Tasks | Creating and working through your to-do list |
| Google Keep | Writing to and pulling from your notes |
| Canva | Starting and editing designs |
| Dropbox | Reaching files stored outside Google Drive |
| Instacart | Building a grocery order |
| OpenTable | Finding and starting restaurant bookings |
| Zillow Rentals | Searching rental listings |
| Google Photos | Finding, enhancing, albuming and sharing your photos in one instruction -- US only, 18+, English, mobile and web only. See below |
| Viator | Finding and booking tours and activities (newly named in July's Gemini Drop) |
| Anything else, via custom MCP | Connecting an app Google has not officially integrated |
| Your own Chrome browser | Chrome auto browse -- driving signed-in websites directly, US only for now |
That last row is the one worth understanding. MCP (Model Context Protocol) is an open standard for plugging apps into AI assistants -- Google is "rolling out support for custom MCP, giving you the ability to connect your favorite apps directly into Tasks in Gemini." In plain terms: you are no longer stuck waiting for Google to strike a deal with a service you use, provided that service offers an MCP connection.
Tasks in Gemini also got a watching job. It can now track topics in real time across blogs, news sites, social media, finance, shopping, weather, and sports -- not just your email. That turns a standing instruction like "tell me if the rent on that listing drops" or "flag anything major about my employer" into something it can genuinely monitor while you get on with your day.
It Can Run Your Google Photos (September 2026)
Tasks in Gemini can now take a single instruction about your photos and carry out the whole chain: find the right shots, enhance them, put them in an album, and share it. Google's Photos help article, re-read live October 16, 2026 and renamed to "Streamline your Google Photos workflows in Gemini Apps", gives the example prompt "Find the best 15 photos from our trip to Hawaii last week. Enhance the beach shots, put them into a shared album called 'Aloha 2026', and draft an email with the link."
This is the most tightly gated thing Tasks in Gemini does, and the gates are narrower than its own. Google's requirements, quoted from that page:
| Requirement | What Google says |
|---|---|
| Age and country | "Be 18 or over and in the US. This feature isn't available in some United States regions." |
| Plan | You must "Have access to tasks in Gemini Apps" -- which since November 17, 2026 includes free accounts in the six countries above |
| Setup | You must "connect Google Photos to Gemini Apps" |
| Language | "Only available in English" |
| Where it runs | "Only available in the Gemini mobile app and on the web at gemini.google.com" -- not the Mac app |
| Rollout | "We're gradually making Google Photos available in tasks in Gemini, so it might not be available to you just yet" |
Note the third row from the bottom. Tasks in Gemini itself runs on the Mac app; the Photos half of it does not. So a US Pro subscriber on a Mac has Tasks in Gemini but cannot use it on their photos except through the browser.
Three things Google promises about how it behaves, which are the ones worth knowing before you point an agent at your photo library:
- Your originals are never touched. "Gemini doesn't edit or overwrite your original photos. To edit a photo, a copy is made and then edited." The copy is saved back to Google Photos.
- New albums start private. "When Gemini creates a new album, it's private to you. Gemini asks for your permission before sharing an album."
- Sharing and emailing need your approval. "Gemini asks for your permission before performing actions like creating shared albums or sending emails."
It can also pull text out of images -- find receipts, extract the text, and summarize it into a Google Doc -- and run recurring photo jobs, like building a monthly highlights album or filing receipt photos automatically. If image generation is the part you care about, the underlying model has its own guide.
What You Are Agreeing To When Gemini Asks Permission
Every permission box Gemini shows you during a task has up to four buttons, and two of them change settings you will never see again unless you go looking. Google documents the whole thing on one page, Manage browser & app permissions for tasks in Gemini, read in full on November 9, 2026. It is the most useful page Google has published about this feature and it is not linked from the announcement. Here is what each button does, in Google's own words:
| What you tap | What it actually does |
|---|---|
| Don't allow | "Declines the action. Gemini stops and won't continue with the action." |
| Allow / Allow once | "Approves the action for this specific step only. Gemini will ask again next time." |
| Allow for this chat | "Approves the action automatically for the rest of the current thread without changing your Connected Apps settings." Google adds that this option "is only available for connected Google products" -- so for a third-party app it is not offered |
| Always allow | "Approves the action for the current thread and all future task threads. Selecting this automatically updates your preference to 'Always allow' in your settings" |
The difference between the middle two and the last one is the whole point. "Allow for this chat" is thread-local and Google tells you how to undo it: "To reset chat permissions, delete the task thread or start a new one." "Always allow" is not thread-local and deleting the thread does not undo it, because it has already written itself into your Connected Apps settings. If you tap it once in a hurry on a Tuesday, Gemini has standing permission for that action forever.
Gemini is supposed to ask in the first place for a specific set of actions, and the list is worth knowing: Google says it asks "for actions that delete data or send messages or emails in Google apps, or handle sensitive third-party tasks, like payments or sharing sensitive info."
Where to Change Your Mind Afterwards
Two separate screens, and most people only find one. App permissions live in Connected Apps; browser permissions live in Browser settings. Both are reachable from the Gemini app and neither is in the main settings list.
| What you want to change | Where | Steps Google publishes |
|---|---|---|
| What Gemini may do inside a connected app | Connected Apps | On a computer: gemini.google.com, sidebar, Customize, then Connectors (click More first if you cannot see Customize), pick the app, then under "App actions" set each action to Always allow or Always ask. On a phone or tablet: tap Menu at the top, and the heading reads "Available actions" rather than "App actions" |
| Whether Gemini asks before opening a browser at all | Settings, then Browser | Next to "Permission", choose Always allow or Always ask |
| Whether it uses your own Chrome or a cloud browser | Settings, then Browser | Under "Preferences": "Let Gemini decide" ("Gemini uses either Chrome on your device or a remote browser for each task depending on availability") or "Remote browser only" ("This works when you're away and your device isn't available"). Google gates this one: "With a Google AI subscription, you can also choose how Gemini browses the web" |
| Delete what the cloud browser kept | Settings, then Browser | Next to "Delete remote browser data", click Delete. On a phone, the button is at the bottom of the screen |
A detail worth not skipping: Google says the Connected Apps setting is the default and the thread choice is the exception. "Your Connected Apps settings set your default ('Always allow' or 'Always ask') across all task threads", and a thread-level "Allow once" or "Allow for this chat" leaves that default alone.
What the Cloud Browser Keeps, and How to Wipe It
If Gemini used the remote browser in Google's cloud rather than your own Chrome, it saved your logins. Google's sentence, verbatim: "When Gemini uses the remote browser to complete tasks, browser data, like cookies that contain your website authentication info (your sign-in details), is saved for future sessions for your convenience." That is the mechanism that lets a scheduled task pick up where it left off without asking you to sign in again, and it is also a store of live session cookies for every site the agent touched.
You can clear all of it at any time, from Settings, then Browser, then Delete remote browser data. Google also says it may do this for you: "To help protect your privacy, Gemini may automatically delete saved browser data on your behalf, and you may need to sign back in to websites." It does not publish how often that happens, or after how long, so treat the manual Delete button as the only timing you control.
One reassurance from the same page that does hold up: a blanket "Always allow" is not absolute. "Gemini is designed to apply built-in safeguards that confirm some actions with you even if you set a permission to 'Always allow'." Google does not list which actions those are, so do not rely on it for anything expensive.
One gap worth flagging, because it matches a pattern. This permissions page lists its own surfaces as "the Gemini mobile app, Gemini web app at gemini.google.com, and the Gemini app on Mac". Windows is absent again -- the same three-surface list, and the same silence, as the tasks help article itself. See the Windows section above.
Is It Safe to Let It Into Your Email and Apps?
Reasonably, if you supervise it -- Google built the obvious guardrails in. You choose whether to turn Tasks in Gemini on at all and which apps it connects to, and it is designed to ask you first before high-stakes actions like spending money or sending emails. That approval step is the single most important safety feature in any AI agent, and you should never turn it off.
The standard cautions still apply, and they are the same ones in our AI agent safety basics: keep an agent out of banking and your primary email unattended, review what it did before trusting the result, and approve every purchase or outgoing message yourself. Agents can also be fooled by malicious web pages ("prompt injection"), so treat Tasks in Gemini like a capable new assistant on their first week -- useful, but watched.
Should You Care Right Now?
If you are in one of the six free countries, yes -- switch the Task toggle on and point it at one annoying recurring chore to see if it earns its keep. Since November 17, 2026 that costs nothing, so there is no upgrade decision to weigh any more, and "watch this space" is no longer the right advice for a free account. The reason to pause is not price and not Personal Intelligence, which Google has taken back off the requirements list: it is the permissions screen. Before you point this at anything that matters, go to Settings, then Browser, and read the two dropdowns there -- they decide whether Gemini asks before it opens a browser as you, and whether it may use your own Chrome. The agent category is still slower and more supervision-hungry than the marketing suggests, which is a reason to start it on something low-stakes rather than a reason to skip it. And if you are in the UK, the EU, Switzerland or Nigeria, the decision is out of your hands entirely.
If you are comparing Google's ecosystem against the alternatives before spending anything, our guides to ChatGPT vs Gemini for everyday use and which AI chatbot is worth paying for cover the tradeoffs. And if the whole "AI that does things for you" category is new to you, start with what an AI agent actually is -- Tasks in Gemini is one of the clearest examples of it.
The bottom line: Tasks in Gemini is a real, working personal AI agent with sensible guardrails, and since November 17, 2026 it is no longer paywalled in the US, Canada, Australia, India, Japan or South Korea. If you are in one of those six and over 18, there is nothing left to decide on price -- open the Gemini app, find the Chat and Task toggle below the prompt box, and the only judgement left is which of your accounts you are willing to connect to it. If you are outside those six it still costs $19.99 a month for Google AI Pro, and in the UK, the EEA, Switzerland and Nigeria it is still not available at any price.
What changed (August 3, 2026): the Daily Brief column carried no country or age gate, and the July 29 note below said Google published none. That was true of Google's subscriptions page and false of Google. Three other vendor surfaces publish the gate: the Daily Brief help article ("Be 18 or over and in the US"), Google's Daily Brief page ("Ultra, Pro, and Plus users over the age of 18 in the United States" -- superseded: as of September 24, 2026 that page's wording is "Availability varies. US-only. English-only. 18+.") and the 2026.05.19 release note ("all Google AI subscribers 18 and over in the United States"). The help article also excludes work, school and supervised accounts outright, so the "Varies by plan" cell for Workspace was wrong too. The summary FAQ, the key takeaway, the availability table and the pricing section were all corrected, and Daily Brief now has its own guide. Spark's own availability is unchanged.
What changed (July 31, 2026): Google removed the US restriction on Google AI Pro. Its July 30 announcement says it is "rolling out access for Google AI Pro subscribers in over 160 additional countries starting today," July's Gemini Drop says Spark is "now available worldwide" (footnote: not in the European Economic Area, United Kingdom, Switzerland, or Nigeria), and the Spark help page now states the requirement as simply "a Google AI Pro or Ultra subscription" with no country attached. This page had said Pro qualified in the US only. The summary, key takeaways, availability table, pricing section, app-connection table and verdict were all corrected, a section on the new Chrome auto browse feature was added, and Viator was added to the connected-app list per the Drop.
What changed (July 29, 2026): Google's Spark help page set out materially wider availability than this page then described. Its archived June 24, 2026 version still required a Google AI Ultra subscription everywhere, ran in English only, worked in the mobile and web apps only, and excluded ten countries including Australia, Canada, India and Japan. The July 29 version admitted US Google AI Pro subscribers, ran in all Gemini Apps languages on Ultra, included the Mac app, and narrowed the exclusions to the European Economic Area, Nigeria, Switzerland and the United Kingdom. The unsourced "(US)" qualifier on the Daily Brief column was also removed: Google's subscriptions page confirms Daily Brief on Plus, Pro and Ultra but publishes no country restriction for it.
Corrections and Updates
November 18, 2026: this page said Tasks in Gemini was Google's paid AI agent and answered "Is Tasks in Gemini free?" with "No". Google made it free in six countries the day before, so this page was telling readers in the US to pay $19.99 a month for something they already had. Four Google surfaces were read live on November 18, 2026 before anything was rewritten, and all four agree. The tasks help article (strip 28,347 characters, against 28,736 on November 16) has lost "Have a Google AI Pro or Ultra subscription" from its requirements list and now reads "Available without a Google AI subscription in Australia, Canada, India, Japan, South Korea, and the US. Available with a Google AI Pro or Ultra subscription everywhere else Gemini Apps are supported, except in the European Economic Area, Nigeria, Switzerland, and the United Kingdom." The announcement, article:published_time 2026-11-17 with no modified time, says "free of charge ... across the U.S., Japan, Korea, Canada, Australia, and India". The release note dated 2026.11.17 says the same. And the feature-availability table now ticks Tasks in Gemini under "Without a Google AI plan".
That last surface is the one worth a warning, because it changed without changing length. 16275805 measured 12,136 characters of stripped text on November 16 and 12,136 today, and in between that cell went from blank to ticked. A checkmark on Google's tables is an <img alt="Checkmark">, so it never reaches the stripped text at all. Any check of this page that compared text length would have reported it unchanged. The cell was re-read from raw markup, where the Tasks in Gemini row carries four checkmark images, before this correction was written.
Corrected here: the title, the summary, the lede, three Key Takeaways, two FAQ answers, the availability table, the requirements block, the "Is Tasks in Gemini Free?" section, the Google Photos plan row, the should-you-use-it section and the closing verdict. Two things were deliberately not written. First, that paid plans no longer matter: Google's models table gives free accounts Hydrogen in task mode and reserves Argon and Carbon for Pro and Ultra, and the four access grades on Google One's plan page have no numbers behind them. Second, that the setting is now called History: the tasks article says History, and Google's skills article still said Keep Activity twice on the same day, so the page names both.
November 9, 2026: this page told you that you must opt in to Personal Intelligence before Gemini will take a task. Google has removed that requirement, so the page was asking you to hand over your Gmail, Contacts, Photos, Search and YouTube history for nothing. Read live on November 9, 2026, the list on the tasks help article under "To assign Gemini tasks, you must:" has four items -- 18 or over, a personal Google Account, a Google AI Pro or Ultra subscription, and Keep Activity on -- and the line "Opt in to Personal Intelligence. Learn more about Personal Intelligence." is gone. The Internet Archive's captures of October 19, October 22 (the en-GB variant) and October 25, 2026 all still carry it, and a word-for-word comparison of the October 19 block against the live one shows those eleven words are the only change: every other requirement, the four-country exclusion and the three-surface device list are identical. So the removal happened between October 25 and November 9, 2026. Google announced it nowhere: the article carries no "Last updated" line, gemini.google/release-notes/ still stops at 2026.10.13 with no mention of tasks, and no blog post or feed entry covers it. Corrected on the summary, the lede, two Key Takeaways, two FAQ answers, the before-and-after table, the availability section and the closing verdict. The October 16 note below is left as written, because it was accurate on the day. A new section on the permission controls was added from Google's permissions article, which this page had never cited.
October 16, 2026: this page described a feature called Gemini Spark in the present tense. Google renamed it to "Tasks in Gemini" in the week of October 13, 2026, and added a requirement: you must now opt in to Personal Intelligence before you can assign a task. The help article is now titled "Manage your tasks & workflows in Gemini Apps" and uses "Spark" twice, both inside a new FAQ reading "Where did Spark go? I can't find it."; the Internet Archive's October 9, 2026 capture (x-archive-orig-date: Fri, 09 Oct 2026 13:10:26 GMT) is still titled "Use Gemini Spark" and uses the word 101 times, which brackets the change to that week. The Gemini Apps Privacy Hub dates it: "Last updated: October 13, 2026", with a new line reading "Learn more about Tasks in Gemini (formerly known as Spark)". Three more Google surfaces had already swapped the name by October 16 (subscriptions, Google One, the Photos article) and one had not (the Gemini Drops carousel still reads "Gemini Spark is available everywhere."). The title, summary, lede, all eight FAQ answers, the Key Takeaways and the closing verdict were rewritten; a new "What Changed Besides the Name" section carries the full diff. No plan, age or country gate moved, and quotations from documents written before the rename keep the words those documents actually use. Also new and now on the page: tasks can hand you finished files (PDF, .docx, .xlsx, .pptx and more), skills are summoned with @ rather than /, and the control moved from a sidebar "Switch to Spark" button to a Chat and Task toggle below the prompt box. Google published no release note for any of it.
September 11, 2026: yesterday this page said flatly that Spark "runs in four places," including the Gemini app for Windows. Google's own documentation does not support that, and the page now shows the disagreement instead of picking a side. The Spark help article, read September 11, 2026, says Spark is "only available in the Gemini mobile app, the Gemini app on Mac, and the Gemini web app at gemini.google.com", and the word Windows does not appear anywhere on it. Two further Google surfaces describe the Windows app without naming Spark: its own help article and the 2026.09.10 release note. Only the launch post says otherwise. Three surfaces to one is not enough to state a flat "yes", and it is not enough to state a flat "no" either, so the key takeaway and the availability section now give the reader both readings and a way to check. Nothing about Spark's plan, age or country gates changed.
September 11, 2026: Google Photos was added. Google's Photos help article was read for the first time on September 11, 2026 and sets out gates this page had never carried: 18+, US only, English only, mobile and web only (not the Mac app), and a gradual rollout. This page had described Spark's Photos connection only from the September 9 Google One roundup, which stated none of those limits. A new section, a connection-table row, a key takeaway and an FAQ answer were added.
September 10, 2026: this page said Spark "runs in three places only." It now runs in four. [Superseded September 11, 2026 -- see the entry above: Google's help documentation still lists three surfaces and does not name Windows. This note rested on the launch post alone.] Google launched the Gemini app for Windows on September 10, 2026 and names Spark as one of the things you can do in it, so the mobile app, the Mac app and the web app are no longer the full list. The key takeaway and the availability section were both corrected. Nothing about Spark's plan, age or country gates changed.
August 13, 2026: Spark now runs on a new model.
August 3, 2026: the Daily Brief column in the table on this page was wrong.
Frequently Asked Questions
Where did Gemini Spark go? I cannot find it in the app.
Google renamed it. It is now called Tasks in Gemini, and you reach it from the Chat and Task toggle below the prompt box instead of a Switch to Spark button in the sidebar. Google says your existing tasks and schedules are still there, in the sidebar.
Is Tasks in Gemini free?
In six countries, yes, since November 17, 2026. Google says it is available without a Google AI subscription in Australia, Canada, India, Japan, South Korea and the US. Everywhere else Gemini Apps run it still needs Google AI Pro or Ultra, and it is blocked outright in the EEA, Nigeria, Switzerland and the UK.
Do I get Tasks in Gemini?
If you are 18 or over, signed in with a personal Google Account and in one of the six free countries, yes, at no cost. Outside them you still need Google AI Pro or Ultra, and in the UK, the EEA, Switzerland and Nigeria you cannot get it at any price. Google also requires your history setting on. Read November 18, 2026.
Did the rename change what the feature does?
No, and Google added to it. The gates, the plans and the blocked countries are identical. What is new is that Gemini can now hand you finished files from a task -- PDFs, Word documents, spreadsheets and slide decks -- and that skills are summoned by typing @ rather than a slash.
What can Tasks in Gemini actually do?
It works in the background across connected apps to handle multi-step jobs: organizing your inbox, setting recurring reminders, watching flight and hotel prices against a budget you set, and pulling research into a file. It asks for your approval before high-stakes steps like spending money or sending an email.
Which apps can it connect to?
Inside Google: Gmail, Google Tasks, Google Keep, and since September 2026 Google Photos. Outside Google: Canva, Dropbox, Instacart, OpenTable, Zillow Rentals, and Viator. Google is also rolling out custom MCP support, an open standard that lets you plug in other apps yourself.
Can it use my Chrome browser?
Yes, if you are in the US. Google switched on Chrome auto browse on July 30, 2026, letting Gemini drive your desktop Chrome using your logged-in accounts and, with permission, your saved passwords. Google says Chrome auto browse is desktop only; the cloud remote browser is used on mobile.
What does "Always allow" actually do?
More than it sounds. "Allow for this chat" approves an action for that thread only. "Always allow" approves it for every future task thread as well, and rewrites that action to "Always allow" in your Connected Apps settings. Only the first is undone by deleting the thread.
How do I stop Gemini using my browser for tasks?
Open gemini.google.com, click Settings, then Browser. Next to "Permission", pick "Always ask" and Gemini has to check with you before it opens a browser for any task. On a Google AI subscription the same screen also lets you choose between your own Chrome and a remote browser.
Does Gemini keep my website logins after a task?
Yes, if it used the cloud remote browser. Google says browser data, including cookies that hold your website sign-in details, is saved for future sessions. You can wipe all of it at Settings, then Browser, then "Delete remote browser data" -- and you will have to sign back in.
Is it safe to let into my email and apps?
Google built in guardrails: you choose whether task mode is on and which apps it connects to, and it pauses to ask before consequential actions like paying or emailing. As with any AI agent, keep it out of banking, review what it does, and approve purchases and messages yourself.