Veo 3.1

What Veo 3.1 is good at

Veo 3.1 is the newest video model in Google's Veo family and adds two things that matter on production pipelines: higher resolution ceilings (1080p and 4K in addition to 720p) and native audio generation. The model takes a prompt and produces footage with dialogue, music, and ambient sound baked in. No separate TTS pass required.

The Fast variant cuts latency at the cost of some fidelity, which is useful when you're iterating on ideas or generating bulk variants. Where Veo 3.1 shines: short product shots, cinematic establishing shots, stylised b-roll for explainers, and single-subject motion from a reference still.

Where it struggles: long-form narrative continuity across shots, precise object placement, and anything requiring frame-level editing. For those, Veo remains one node in a larger pipeline rather than the whole pipeline.

Resolution, duration, and aspect-ratio rules

The Veo 3.1 API imposes a handful of constraints that matter when you're wiring a flow. 720p accepts 4s, 6s, or 8s durations in either 16:9 or 9:16. 1080p locks the clip to 8s and 16:9 only. 4K locks to 8s and 16:9. Passing a reference image also forces the clip to 8s regardless of resolution.

The Video node in PlugNode encodes these rules in the config panel. Changing the resolution updates the duration and aspect-ratio options automatically so you can't save a flow that will fail at execution time.

If you need a 9:16 vertical clip at high resolution for Reels or TikTok, the current answer is 720p. Veo has not shipped 4K vertical at the time of writing.

Native audio and what it includes

Veo 3.1 generates audio alongside the video frames, synchronised to the visual. That audio includes dialogue when the prompt describes speakers, ambient sound that matches the scene (wind, crowds, room tone), and stylised music cues when the prompt specifies them.

It is not a replacement for a mastered soundtrack and it will not match a specific voice from a prompt. For branded voiceover you still want ElevenLabs on top.

A typical pipeline on PlugNode looks like: Gemini Text generates the script, Video (Veo 3.1) renders the shot with native ambient audio, ElevenLabs Audio generates a branded voiceover, and an output node returns both streams. Because the whole thing is one flow, re-running with a tweaked script takes one click instead of four tools.

Running Veo 3.1 as an API from PlugNode

This is the wedge. Drop an HTTP Trigger node on the canvas, wire it into a Video node configured for Veo 3.1, and chain a Respond to Webhook node at the end. Hit Publish. You now have a rate-limited, secret-protected HTTP endpoint that accepts a prompt and returns a video URL.

Your product code POSTs to that URL; the flow runs on the server engine; the MP4 lands back as JSON. The endpoint is versioned. Every publish creates a hash-diffed snapshot you can roll back to if a new prompt format breaks downstream.

Secrets rotate from Settings, rate limits are 60 requests per minute per trigger, and the response is either synchronous (pass ?wait=true) or asynchronous with a webhook callback. That is what "publish a flow as an API" means in practice. See the publish-as-API pillar for the full pattern, the Gemini Video integration for the node reference, and the HTTP Trigger integration for trigger-side details.

Where Veo 3.1 fits next to Veo 3 and Veo 2

Veo 2 is 720p only and does not generate audio. It remains cheaper per second and is a sensible default when you need bulk low-res variants or when audio would be stripped anyway. Veo 3 adds 1080p and native audio and is the balance option. Veo 3.1 adds 4K, longer video extension, and refinements in motion stability at higher resolutions.

In PlugNode, switching between the three is a dropdown change on the Video node. Everything downstream stays the same.

A production pattern that works well: draft with Veo 2 or Veo 3.1 Fast while iterating on prompt and script, then swap to Veo 3.1 at 1080p or 4K for the final publish. No other downstream node needs to change.