Is Gemini Deep Research free, and how many reports do you get per plan?
Yes, it is free to run. Google's help page says anyone 18 or over and signed in to the Gemini app can use it, and all users can use the Thinking model. What Google no longer gives is a number of reports. Its model access page, read on 15 November 2026, says "Premium models and features require more usage and may cause you to reach your limit faster", and it lists Deep Research under them.

| Plan | Usage limit (Google's model access page) | Report count printed |
|---|---|---|
| No Google AI plan | Standard limits | None |
| Google AI Plus | 2x higher than standard | None |
| Google AI Pro | 4x higher than standard | None |
| Google AI Ultra | 5x or 20x higher than AI Pro, depending on subscription | None |
The meter is compute-based. Google says it "will refresh every 5 hours until you reach your weekly limit" and that the changes began on 17 May 2026 for users over 18. A research report is a heavy request, so it drains the meter faster than a chat, and nobody outside Google can say how many fit in your week. Treat any guide that gives a fixed daily or monthly count as describing the older limits page. The Gemini Deep Research limit page keeps the dated history of those caps and the counts work accounts still get.
How do you use Gemini Deep Research?
Open gemini.google.com on a computer, click Add Files in the text box, and choose Deep Research. Google's help page for the feature, read on 15 November 2026, gives these steps in order.
- Sign in to the Gemini app. Google requires you to be 18 or over.
- Optional: click Add Files, then Files, to upload files or an image with your prompt.
- Optional: click Sources and pick what to include, such as Gmail or Drive. Those two appear only after the Google Workspace app is connected to Gemini Apps.
- Type what you want researched and click Submit.
- Gemini writes a research plan. Click Edit plan to change it first.
- Click Start research. You can leave the chat, and Gemini notifies you when it finishes.
- When the report is ready, click Open.
Google Search is the default source, and the help page puts it plainly: "Google Search is included as a source by default." Deselect it and the report draws only on the files, Gmail, Drive or notebooks you chose. You can also add NotebookLM notebooks, which the help page lists next to files. Google's mobile steps match (tap Add Files, then Deep Research), though source choice is still rolling out gradually in the mobile app. A finished report arrives as a notification on the device, even when it is locked, depending on your settings.
What does Gemini Deep Research actually do with your question?
It turns the question into a plan, searches, reasons over what it finds, and writes a report. Google's Deep Research page names those four stages, and says the feature can "automatically browse hundreds of websites" and, if you choose, your Gmail, Drive and Chat. It shows its thinking as it goes, so you can read what it has learned and what it plans to check next.
The plan is where you steer. Google's own description is that Gemini presents the plan and you "can refine it to make sure it's focused on the right areas." In practice that means deleting a branch you don't care about, adding a region or a date range, and telling it which kinds of source to trust. Do it before you click Start research. A plan edit is free. A rerun spends a report from your allowance.
The report comes back as a multi-page document with each source linked as a citation, per Google's March 2025 Workspace post, which also describes the follow-up loop: ask a question, request a change, or have Gemini search a subtopic. Google says the Gemini 3 model made Deep Research better at every stage, "from planning to delivering even more insightful and detailed reports." The same page lists 150 countries and 45+ languages, and it carries no date.
What prompts work well in Gemini Deep Research?
Prompts that name the decision, the sources and the output shape work best, because the plan is built from them. These five are written for this page, one per job. Swap the bracketed parts for your own.
- Competitor teardown: "Build a one-page brief on how [competitor] prices and packages [product] compared with ours. Use public pricing pages and customer reviews from the last 12 months. End with three risks for us and one move we could make next quarter."
- Meeting prep: "I'm meeting [company] on Friday. Summarize its products, funding history, leadership and main rivals, with a source for every claim. Put anything you could not confirm in a separate list called Unverified."
- Literature scan: "Find the five most cited papers on [topic] since [year]. For each, give the question, the method, the main result and one limitation the authors admit. Then say where the field disagrees."
- Claim check: "I read that [claim]. Find the original source, say what it actually shows, and list the best evidence for and against. If the claim traces back to one study or one press release, say so."
- Buying decision: "Compare the three best-reviewed [product type] under [budget] on [two or three features that matter]. Use professional reviews and owner complaints from the last 12 months, and give a pick for [use case] with the one reason it could be wrong."
Before you click Start research, read the plan and fix one thing. If it lists a source type you don't trust, remove it. If it misses the date range, add it. If a prompt needs your own notes, upload the file or turn on the Drive source first, since the plan can only use what you attach. The Unverified list in prompt 2 and the one-reason-it-could-be-wrong line in prompt 5 are cheap ways to make the report show its weak spots.
Which plan and sources does Gemini Deep Research need?
No paid plan is required to run a report, but the plan decides the model, the limits and the visuals. As of 15 November 2026, Google's help page sets these conditions.
| What you get | Who gets it | What Google's help page says |
|---|---|---|
| A Deep Research report | Anyone 18 or over, signed in to Gemini | Be 18 or over, be signed in |
| Thinking model | All users | "All users can use Thinking for their reports" |
| Pro model | Google AI Pro and Google AI Ultra | Pro "for even higher quality" |
| Higher report limits | Google AI Pro and Google AI Ultra | Limits on daily requests and on requests running at once |
| Charts, diagrams, interactive simulators | Google AI Ultra | Not available when Gmail or Drive is a source |
| Gmail and Drive as sources | Anyone who connected the Workspace app | Off until connected |
Here's the catch. An Ultra report can include charts and simulators, and the moment you tick Gmail or Drive, Google says the visuals stop. So pick one: your own mail and files as evidence, or the charts. For a report that needs both, run the web report with visuals first, then ask a follow-up in the same chat that uses your files.
The help page gives no report count. It says Gemini tells you how many requests are left when you are close to the day's limit, and that Pro and Ultra get more. The multipliers by plan are in the table at the top of this page.
What is Gemini Deep Research Max?
Deep Research Max is the more thorough of two Deep Research agents Google released for the Gemini API on 21 April 2026. Both run on Gemini 3.1 Pro. Google says the standard agent is "tuned for speed and efficiency" for interactive use, while Max uses extended test-time compute to reach "maximum comprehensiveness and highest-quality synthesis", and suits background jobs such as an overnight report.
| Deep Research | Deep Research Max | |
|---|---|---|
| Built for | Fast, interactive use | Thorough background runs, such as overnight reports |
| Model | Gemini 3.1 Pro | Gemini 3.1 Pro |
| Where you reach it | Gemini API, Interactions API | Gemini API, Interactions API |
| Status | Public preview, paid API tiers | Public preview, paid API tiers |
| Sources | Web search, remote MCP servers, file uploads, file stores | Same |
| Output | Charts and infographics inline, editable plan | Same |
The practical point is that Max is not a toggle in the Gemini app. Google's announcement does not list it as an app option, and the app help page does not mention it. If you open the app and run Deep Research, you are using the app feature and its plan meter. Max is for developers and teams who call the API on a paid tier and want the extra depth. Google also says it is coming to startups and enterprises on Google Cloud.
How do you export a Gemini Deep Research report to Google Docs?
Open the report chat, find the Canvas panel on the right, and click Share & export, then Export to Docs. Google's help page lists three choices under that button: Share Canvas to share it, Export to Docs to export it, and Copy Contents to copy the text.
- Go to gemini.google.com. If your chats are hidden, click Menu at the top.
- Under Recent, select the chat with your research report.
- In the Canvas panel on the right, click Share & export.
- Click Export to Docs.
Past reports only show up in Recent when your Keep Activity setting is on, so a missing report is usually that setting. Two other Canvas buttons are worth a click. Create, then Audio Overview, turns the report into a listenable summary. Create, with a description typed in the text box, builds a custom visualization from the report.
Gemini vs ChatGPT vs Perplexity deep research: which one?
None wins every job. Gemini exports straight to Google Docs and can read Gmail, Drive and Chat. ChatGPT lets you restrict research to chosen websites and exports to Markdown, Word and PDF. Perplexity is the fastest on paper and the only one of the three that still prints a free count. The table is from each vendor's own pages, read on 15 November 2026.
| Feature | Gemini Deep Research | ChatGPT deep research | Perplexity Deep Research |
|---|---|---|---|
| Where you start | Add Files, then Deep Research | Type /Deepresearch, use the + menu, or the sidebar | Deep Research in the mode selector of the search box |
| Time the vendor states | Usually about 5 to 10 minutes | 5 to 30 minutes | 2 to 4 minutes (launch post) |
| Plan you can edit first | Yes, Edit plan | Yes, review and modify the plan | Not stated |
| Personal sources | Gmail, Drive, Chat, NotebookLM notebooks | Connected apps such as Google Drive or SharePoint, where your plan allows | Not stated |
| Limit it to chosen sites | Not listed | Yes, Sites then Manage sites | Not stated |
| Export | Google Docs, copy text, share Canvas | Markdown, Word, PDF | PDF, document or Perplexity Page |
| Free access | Yes, standard limits, no count | Varies by plan; a counter shows remaining tasks | 1 research query a month on the plan comparison page |
| Paid allowance | 2x, 4x, 5x or 20x multipliers, no count | Varies by plan, fixed monthly allowances reset 30 days from first use | Monthly limits on Pro and Max, no count |
Pick Gemini if your work lives in Google Docs, Gmail and Drive, or you want an Audio Overview of the result. Pick ChatGPT if you need to pin research to a short list of trusted sites or hand a Word or PDF file to someone outside Google. Pick Perplexity for a quick source-first answer when one report a month on the free plan is enough. Our ChatGPT deep research guide covers that side step by step.
Whizi, which publishes this page, does not run Gemini Deep Research, because that is Google's own feature. If you want to put the same question to Gemini, GPT and Claude and compare the answers, Whizi Pro carries all three in one app, with web search on by default and a first week for $1.99. The free plan gives 10 messages a day on Whizi AI, Auto and Claude Haiku 5.5, with no card. It suits checking a report against a second model, not replacing the report.
When Gemini Deep Research fails, what do you do?
Check four things in order. Google's help page names the conditions, and each one fails quietly.
- Age and sign-in. You must be 18 or over and signed in to the Gemini app.
- Limits. There are caps on daily research requests and on how many run at once. Check that no other report is still running.
- Sources. Gmail and Drive appear only after the Workspace app is connected, and visuals vanish when either is a source.
- History. If a finished report is missing, turn Keep Activity on, then look under Recent.
| What you see | Likely cause | What to do (our advice, not Google's) |
|---|---|---|
| Stuck on starting research | A report already running, or the limit | Wait for the running one, then check the request count Gemini shows near the limit |
| Limit message before you expected it | Reports spend the compute meter faster than chats | Wait for the 5 hour refresh, or move to a plan with a higher multiplier |
| Report is thin or generic | The plan was too broad | Rerun with a narrower question, a date range and named source types (our advice) |
| A cited link does not support the claim | Any research agent can misread a source | Open the link before you rely on the number (our advice) |
| Report missing from Recent | Keep Activity is off | Turn it on, then look again |
| Gmail or Drive ignored | Workspace app not connected | Connect it, then choose it under Sources |
If all of that is fine, start a new chat and run the report again. That is our advice, not Google's, and it works because a fresh chat carries no earlier thread into the plan. Developers have another door: Google announced Deep Research and Deep Research Max for the Gemini API on 21 April 2026, built on Gemini 3.1 Pro and in public preview through paid tiers. Deep Research is tuned for speed and Max for comprehensiveness, and both are reached through the Interactions API, not generate_content.
- Start from Add Files, then Deep Research, on gemini.google.com while signed in and 18 or over
- Click Edit plan and fix the sources and the date range before you click Start research
- Leave Gmail and Drive off if you need Ultra charts, since Google says visuals are not available with them
- Export from the Canvas panel: Share & export, then Export to Docs
- Turn Keep Activity on, or finished reports won't appear under Recent
- Name the decision and the output shape in the prompt, and ask for a source on every claim
Frequently asked questions
Is Gemini Deep Research free?
Yes, any signed-in user 18 or over can run a report, and all users can use the Thinking model. Google prints no report count. Its model access page gives standard limits without a plan, 2x on AI Plus, 4x on AI Pro and 5x or 20x AI Pro on Ultra. Gemini Deep Research limit covers the history.
Can Gemini Deep Research export to Google Docs?
Yes. In the Canvas panel of the report chat, click Share & export, then Export to Docs. The same menu has Share Canvas and Copy Contents. Google's March 2025 Workspace post described the same export as the last step of the flow, so it has been part of the feature since launch.
Can Gemini Deep Research read my Gmail and Drive?
Yes, if you connect the Google Workspace app to Gemini Apps and choose them under Sources. Google Search stays on by default unless you deselect it. The cost is the visuals: Google says charts and simulators in Ultra reports are not available when Gmail or Drive is a source.
How long does a Gemini Deep Research report take?
About 5 to 10 minutes, and longer for complex reports, per Google's help page. You can leave the chat while it runs. On the web app the finished report is flagged next to the chat thread, and on mobile you get a notification.
What is Gemini Deep Research Max?
Deep Research Max is the more thorough of two Deep Research agents Google released for the Gemini API on 21 April 2026, both built on Gemini 3.1 Pro and in public preview. Google says the standard agent is optimized for speed and Max for comprehensiveness. Google's announcement does not list Max as a Gemini app option. Both run through the Interactions API.