Can Gemini watch videos?
Last updated 2026-10-20
The short answer
Yes, within limits. Google's Gemini Apps Help says you can upload videos, up to 10 files per prompt and 2 GB each, with 5 minutes of total video, or 1 hour with Google AI Pro or Ultra. It can also help you understand a YouTube video. That gets you feedback, not a finished cut: understanding a video is not editing it. Woven Video, a Mac editor, is one way to get edits from an assistant that sees your frames.
Your options
Upload the file in the Gemini app. Google's upload page lists videos among the files you can add on the web app: up to 10 files per prompt, each video up to 2 GB, and total video length up to 5 minutes, extended to 1 hour with Google AI Pro or Google AI Ultra. If Gemini reports that it failed to analyze a file, the page says to upload it again.
Ask about a YouTube video. Google's page on YouTube content says Gemini uses public information from YouTube to help you understand the content of a video. You need to be signed in with Keep Activity on, and the feature isn't available in Live chats or in Gemini in Google Messages.
Cut it down first. A 12-minute recording is over the 5-minute total on the standard limit, so trim it to the part you want judged, such as the opening 30 seconds.
Developers: use the Gemini API. Google's developer docs describe sending a video through the Files API, inline for small files, or as a public YouTube URL, which they label a preview. They say video is sampled at one frame per second by default and that fast motion can lose detail. Those docs describe the API; the app's help page doesn't say how the app reads a video.
Another way to get frame-aware edits on a Mac
Woven Video is an AI video editor for Mac. Its agent reads still frames of your footage and the speech as timed text, then makes the changes on a timeline.
- What it sees. Stills of your footage and of the finished reel, the speech as text with word timings, and the matched song. Google's API docs say Gemini models process audio and visual streams; Woven's agent gets neither motion nor audio playback and is meant to say what it couldn't check.
- Edits, not only notes. Ask in chat for the cut and the agent trims, orders clips, adds captions, music, sound effects and effects on a multi-track timeline. You can change any of it by hand, then export an MP4.
- Study a video you like. Give the in-app agent a link and it downloads the video, works through its frames and speech, matches the song and saves a written breakdown of the hook, pacing, shots and sound. Videos longer than 90 seconds are broken down for their first 90 seconds only.
- Which AI apps connect. Settings, then AI apps, offers Claude, Codex (ChatGPT) and Other app, with no Gemini entry. Woven's own chat runs on Woven-hosted models billed to credits, or on a paid ChatGPT plan.
- Mac only. Woven runs on macOS 15 or later, and it needs the internet for sign-in, chat, transcription and downloads.
Pick by the question. If you only need a summary of a video, Gemini may be enough. If the cut itself has to change, use an editor with a timeline.
Get frame-aware edits in Woven
- 1Sign in. Open Woven on your Mac and sign in with Google.
- 2Bring in footage. Click Add, choose Video, or drag clips in from Finder. MP4, MOV, M4V and WebM all import.
- 3Ask what is in it. Ask the chat what each shot shows and what is said over it, then describe the cut you want.
- 4Check and export. Play the edit back and fix anything by hand on the timeline. When it looks right, export it as a 1080p or 1440p MP4 from the Export button.
Woven needs macOS 15 or later and a signed-in account on a plan or the 3-day free trial. Transcription and captions are paid from Woven credits.
FAQ
Yes in the Gemini app, within the limits above. Google's help page doesn't say how the app reads the video, so treat details such as fast motion or exact timing with care. Understanding a video is separate from editing it.
Google's help says Gemini uses public information from YouTube to help you understand a video's content. You need to be signed in with Keep Activity on, and it isn't available in Live chats. The developer docs describe public YouTube URLs as a preview feature and exclude private and unlisted videos.
It is the name of Google's developer docs page on video input for the Gemini API. The page says Gemini models can describe, segment and extract information from videos, answer questions about content and refer to timestamps written as MM:SS.
Yes. Google's help lists videos among the supported files in the Gemini web app, with the limits above. The API docs list MP4, MPEG, MOV, AVI, FLV, MPG, WebM, WMV and 3GPP as supported video formats.
In a limited way. Google's help on generating videos lists prompt edits such as removing or replacing objects, and a video-to-video mode where you upload a video and ask for edits. It needs a Google AI plan, and uploading videos for edits isn't available in the EEA, Switzerland, the UK or some US states. For cuts, order, captions and music on a timeline, you need an editor.
Woven's Settings, under AI apps, lists Claude, Codex (ChatGPT) and Other app, with no Gemini entry. Woven's own chat runs on Woven-hosted models or a paid ChatGPT plan, so you don't need Gemini to edit in it.
More answers
Want the editing side, or another assistant's answer? See our answers for Can Gemini edit videos?, Can ChatGPT watch videos? and Can AI edit my video?.
Sources
- Upload & analyze files in Gemini Apps (Gemini Apps Help), checked October 20, 2026
- Find & ask about YouTube content in Gemini Apps (Gemini Apps Help), checked October 20, 2026
- Video understanding (Gemini API, Google AI for Developers), checked October 20, 2026
- Generate videos with Gemini Apps (Gemini Apps Help), checked October 20, 2026
Try Woven free for 3 days
Then $25/mo billed yearly ($299/yr) or $39/month. Native Mac app for short-form video.
Paste into Claude Code to install and connect Woven.