Can Gemini watch YouTube videos?

Yes, for public videos. Paste a YouTube link into Gemini and it can summarize the video and answer questions about it; Google's developer docs say the model takes in both the sound and the picture, one frame per second. What the Gemini app does with a video that has no captions is not documented, and the guides that rank on this question disagree.

Below: the steps, what Gemini actually reads, where it breaks, and how to get the exact words.

How do you summarize a YouTube video with Gemini?

Google's help sets two conditions. You must be signed in to the Gemini app, and Keep Activity must be on: “This feature isn't available when this setting is off.” It also doesn't work in Live chats or in Gemini in Google Messages.

  1. Go to gemini.google.com, or open the Gemini app, and sign in.
  2. Check that Keep Activity is on in your Gemini Apps activity settings.
  3. On YouTube, copy the video's link from the address bar or the Share button.
  4. Paste the link into the chat with a short prompt, such as “Summarize this video.”
  5. Ask follow-ups in the same chat: a point to expand, where a topic comes up, a quiz.

TechRadar's reviewer asked “When in the video does it discuss restaurants in Tokyo?” and got a couple of sentences plus a link to where that part starts.

Older guides tell you to switch on a “YouTube extension” first, under Settings, then Extensions. Google's help now calls these Connected Apps, found under Customize in the sidebar. It also says Gemini uses public YouTube information automatically, as long as Keep Activity is on.

Does Gemini really watch the video, or read the captions?

It depends on the route.

Through the Gemini API

Google's video understanding docs, updated October 13, 2026, are explicit. Gemini processes “both the audio and visual streams.” By default it samples one frame per second and takes the audio at 1 Kbps. A public YouTube link can go straight into the request. On this route the model gets the sound itself, so it doesn't depend on captions.

A developer's log on Google's forum shows both at work: one request on a video under three minutes counted 44,184 video tokens and 5,351 audio tokens. Some newer models also offer an opt-in “agentic” mode that inspects the transcript first and loads only the visual segments it needs.

In the Gemini app

Google's help only says Gemini “uses public information from YouTube” to help you understand a video. It doesn't say whether that means captions, audio or frames.

The tests point both ways. In May 2026, ZDNet gave the Gemini web app, on the $20-a-month Pro plan, a YouTube link and two large video files. Gemini handled all three, and described a silent drone clip from the picture alone. That clip was an uploaded file, though, not a YouTube link.

Lynote's June 2026 guide says the reverse: Gemini “cannot ‘see’ the video.” Its FAQ goes further: without closed captions, “Gemini cannot process the URL.” A 2024 guide on iQOO's community site said the same thing, citing Google: not every video can be summarized “due to the necessity of captions.” Neither squares with what Google documents for the API.

So for the app, the honest answer is to test it on your own video. One quick check: ask what is shown on screen at a minute where nobody speaks. An answer built from captions can't know. The full five-step test is in What AI can watch YouTube videos?

Where does Gemini break on YouTube videos?

Can Gemini transcribe a YouTube video?

It will write one if you ask. But that text is generated by the model, not copied from the captions, and the forum reports above show it can come from the wrong video. Use it for the gist, not for quotes.

When the video has captions, the free and exact route is YouTube's own transcript. Click Show transcript under the description, copy the text, and paste it into Gemini. Lynote calls this the most reliable method, because Gemini then “analyzes the exact words spoken.” If the button is missing, see YouTube transcript not showing.

When the video has no captions, that panel stays empty. And what the app does then is the part Google doesn't document.

No captions, and you need every word, not a summary? That's the case I built Transcript Any Video for. You paste the YouTube link, and it transcribes the video's audio, not its captions, so it works on videos that have none. You get the transcript word for word, with just your email, no password. Videos can run up to 2 hours. You buy credits once, 10 for $1.99, no subscription, and you can ask for a refund. Then paste the text into Gemini, ChatGPT or Claude and ask your questions.

Transcribe a video

Is YouTube's Ask button the same thing as Gemini?

Not quite. YouTube's help calls Ask YouTube on the watch page “a different feature and experience from Gemini Apps, although both are built with Gemini large language models.” It shows up on select videos, in select languages and regions, so you can ask without leaving the video. Its answers “draw on info from YouTube and the web,” and YouTube warns that quality and accuracy may vary.

Working in another assistant? See Can ChatGPT watch YouTube videos? and Can Claude watch YouTube videos?

What do the top results leave out?

The first result today is a February 2024 video tutorial, then a Reddit thread, then TechRadar's July 2025 piece, which says Gemini does more than transcribe or guess from the title, without saying how. Of the ranking pages I could read, none separates the app from the API or mentions Keep Activity. The step-by-step guides still point to an extension switch Google now files under Connected Apps, and repeat a “no captions, no summary” rule that Google's developer docs contradict.

FAQ

Does Gemini work on unlisted or private YouTube videos?

Not through the API. Google's developer docs say you can only pass public videos, “not private or unlisted videos.” On Google's developer forum, one builder puts it plainly: Gemini can't read an unlisted, private or password-protected video because it has no access to it. The app's help only speaks of “public information from YouTube.”

Does using Gemini on YouTube videos cost anything?

In the app, Google's help asks only that you be signed in with Keep Activity on, and names no paid plan. For developers, passing a YouTube link is a preview feature “available at no charge,” capped at 8 hours of YouTube video per day on the free tier, with no length limit on the paid tier. Google adds that pricing is likely to change.

How long a video can Gemini handle?

Through the API, models with a 1 million token context window take videos up to 3 hours at low media resolution, or 1 hour at high resolution. For a video file uploaded to the app, total length is capped at 5 minutes, or 1 hour on Google AI Pro and Ultra, with each file up to 2 GB.

Can you ask Gemini about one specific minute?

Yes. Google's developer docs say to write the moment as MM:SS, for example 01:15 for one minute and fifteen seconds. In the app, a plain question works too, such as asking when a topic comes up. Check the answer against the video before you quote it.

Sources