The Veo3 8 Second Limit: Why Every Clip Stops at 8s — and How to Make Longer Videos

What is the maximum video length on Google Veo 3?

Eight seconds, full stop. Both Veo 3 and Veo 3.1 generate clips of at most 8 seconds per request, and the duration selector offers exactly three options: 4, 6, or 8 seconds. This is the model's documented per-generation limit as of 2026 — not a restriction invented by any single app. You hit the same ceiling in Google Flow, in the Gemini app, through the Vertex AI API, and on every third-party platform that runs Veo, including ours.

That also means the searches for a hidden switch come up empty by design. Upgrading from Google's AI Pro plan to the $249.99/month AI Ultra plan buys you more generations per day, not longer ones. The 8-second cap is baked into how the model is served, so the honest framing is: you don't bypass the limit, you build around it— and with Veo 3.1's image-to-video continuation, building around it works far better than it used to.

How to make videos longer than 8 seconds on Veo 3

The workaround is clip chaining: generate a sequence of clips that are designed to join, then assemble them. Done carelessly it looks like a slideshow; done with frame continuation it reads as one continuous video. Here is the full workflow:

1. 1

   Split the story into 8-second scenes

   Write your video as a shot list where each beat fits inside 8 seconds — a 30-second ad becomes four beats, a 60-second explainer becomes eight. Plan each cut point at a natural scene change (a camera move, a new angle, a subject turning) so the joins land where a human editor would cut anyway. This planning step is where most of the final quality is decided.
2. 2

   Generate each segment with consistent prompts

   Keep a fixed "style block" — the same wording for characters, wardrobe, lighting, lens, and color grade — and reuse it verbatim in every prompt, changing only the action. Prompt drift is the number-one cause of visible seams: if clip one says "golden hour, 35mm, warm tones" then every clip must say it too.
3. 3

   Chain clips with Veo 3.1 frame continuation

   This is what makes joins nearly invisible. Export the final frame of clip one and use it as the starting image for clip two via image-to-video generation — Veo 3.1 continues the motion from exactly that frame, so subject position, lighting, and composition carry over instead of resetting. Repeat down the chain: every clip inherits its opening frame from the previous clip's close.
4. 4

   Assemble and polish the final video

   Drop the clips onto one timeline in CapCut, DaVinci Resolve, or Premiere Pro (or concatenate losslessly with FFmpeg). With good frame continuation a straight cut usually works; where a seam still shows, a 2–4 frame crossfade hides it. Finish with a single color-grade pass across the whole timeline and level the audio so volume doesn't jump between segments.

Duration options and what each clip costs

Since longer videos are assembled from multiple generations, the per-clip price is what determines what a 30- or 60-second video really costs. On VEO3 Gen, clip duration scales the credit price directly — 4-second clips cost half the 8-second rate and 6-second clips three quarters:

Stack those into real project math: a 30-second video is four chained clips (8+8+8+6s) — 38 credits on Fast, roughly $2–3 at published credit-pack prices, or just 12 credits (well under a dollar) on Veo 3.1 Lite for drafts. A 60-second videois eight 8-second clips: 80 credits on Fast, 24 on Lite. Credits are valid at least 30 days from purchase (see Terms), and there's no daily generation cap — which matters, because chaining is iterative and you'll often re-roll a middle segment two or three times. For the full plan breakdown see our Veo 3.1 pricing guide; if you're currently rationing Google Flow's daily allowance, the Flow credits comparison shows why per-clip pricing suits multi-clip projects better.

Tools for stitching Veo 3 clips together

Any editor that can put clips on a timeline can assemble a chained Veo video. Pick by skill level and how much finishing control you want:

For most creators the practical pairing is Lite or Fast generations for iteration, DaVinci Resolve or CapCut for assembly, and one shared grade across the timeline. Developers automating longer videos can script the whole loop — generate, extract last frame, continue, concatenate — against the API using FFmpeg for the joins; the API docs cover image-to-video requests.

Will the 8-second limit be raised?

Not yet, and nothing official says when. As of mid-2026 Google has not announced any increase to the per-generation cap for Veo 3 or Veo 3.1, and Veo 4 has not been officially announced or released— treat any page promising "Veo 4 with 60-second generations" as speculation. The realistic expectation is that limits will lengthen gradually as models and serving hardware improve, the way image models grew from low to high resolution over several generations.

In the meantime, chaining isn't a stopgap to be embarrassed about — it's how professional video already works. A typical 30-second commercial contains six to twelve distinct shots; nobody films it as one take. Thinking in 8-second shots pushes you toward the same discipline: tighter scenes, intentional cuts, and a script that earns every second. The limit shapes the craft; it doesn't cap the runtime.