Veo 3.1

7 models available

Provider: Google

Updated: Recently

Google DeepMind's upgraded AI video model offers realistic motion generation, extended video duration, multi-image reference control, and synchronized native audio output, supporting 1080p image quality.

Available Models

Veo 3.1 Lite 4K

veo3-1-lite-4k

Veo 3.1 Lite's 4K-supported model.

Veo 3.1 Lite

veo3-1-lite

Veo 3.1 Lite is an efficient developer-first video model that delivers high-fidelity video generation, editing, and cinematic-level control capabilities. Leveraging the advanced Veo 3.1 model, it provides scalable programmatic interfaces for creators and enterprises, thereby democratizing professional-grade video AI.

Veo 3.1 Quality 4K

veo3-1-quality-4k

Veo 3.1 Quality's 4K version.

Veo 3.1 Fast 4K

veo3-1-fast-4k

Veo 3.1 Fast's 4K version

Veo 3.1 Fast Reference

veo3-1-fast-ref

You can provide different reference images (up to 3) to shape character design, lighting style, or color tone, ensuring that the generated video maintains visual consistency in each shot.

Veo 3.1 Quality

veo3-1-quality

Create high-quality, 8-second videos with sound using Gooles's state-of-the-art video generation model.

Veo 3.1 Fast

veo3-1-fast

Veo 3.1 Fast for Generation transforms creative ideas into compelling video narratives using Google's advanced video generation model. Veo is capable of generating videos with audio from text prompts, or animating images with textual guidance.

VEO 3.1 · Google DeepMindNew · Extend to 15s

Eight seconds
is no longer
the end.

Veo 3.1 renders 8-second cinematic shots with native sound. Now hand any take you love to Extend, and the same world, in the same light, keeps rolling to second fifteen.

Start creating API docs

Source · veo3-1-fast

16:9 · 720p · 24 fps

Length: 8s → 15s
Resolution: 720p · 1080p · 4K
Aspect: 16:9 · 9:16
Sound: Native audio
Tiers: Quality · Fast · Lite

Extend00:00:08:00 → 00:00:15:0001 / 07

Keep the camera rolling.

Point veo3-1-fast-extend or veo3-1-quality-extend at a finished take and write what happens next. Veo picks up from the last frame at second eight, so camera, light and characters carry straight through the cut.

Sequence 01 · 24 fps · 00:00:15:00

Returns

V1Source

veo3-1-fast · 8s

veo3-1-fast-extend · +7s

The camera keeps diving along the rain-soaked overpass as a hover car skims past the neon billboards

OUTOutput

15s MP4 · source + extension

POST /v1/videos/generationsmodel veo3-1-fast-extendremix_task_id 7f3a…c21eraw false

01
Shoot the source
Generate with veo3-1-fast or veo3-1-quality as usual and keep the returned task_id.
02
Submit the extension
Switch to the matching extend model, pass the task_id as remix_task_id and describe what comes next.
03
Collect the cut
Poll the new task for the 15-second video, or set raw to true to receive only the new segment.

Editor's notes

Same family only: the fast line (including 4K and ref) uses veo3-1-fast-extend, the quality line uses veo3-1-quality-extend. Lite is not supported yet.
The source must be a completed task from your own account, created within the last 48 hours.
An extended video can't be extended again; each source tops out at 15 seconds.
The aspect ratio follows the source unless you set 16:9 or 9:16. Resolution can be 720p, 1080p or 4K.
Extension works from the source alone: no new images, video or audio, and no duration to set.

Line-upREEL 01 — 0702 / 07

One family, every job covered.

From rough exploration to final delivery, pick a tier by quality, speed and budget. The table also shows which extend model pairs with each tier.

Model	Best for	Input	Resolution	Extend
veo3-1-quality	Flagship quality for final delivery	Text · first/last frame (≤2)	720p · 1080p · 4K	veo3-1-quality-extend
veo3-1-fast	The balance of speed and cost	Text · first/last frame (≤2)	720p · 1080p · 4K	veo3-1-fast-extend
veo3-1-fast-ref	Multi-image reference for consistency	Text + references (≤3)	720p · 1080p · 4K	veo3-1-fast-extend
veo3-1-lite	Lowest cost for bulk exploration	Text · image	720p · 1080p · 4K	Not yet
veo3-1-*-4k	Same tier, locked to 4K output	Same as its tier	4K only	fast-4k / quality-4k can extend

SoundA1 · 48 kHz03 / 07

veo3-1 · sound on

Waveform measured from this clip's actual audio track

Pictures that bring their own sound.

Veo 3.1 generates audio alongside the picture: a fingertip scratching across paper, the wet slap of tentacles on wood, every sound locked to its frame with nothing to dub afterwards.

Foley: Actions and materials sound the way they look.
Sync & rhythm: Dialogue, footsteps and cuts stay in time.
Ambience: Rain, crowds and city hum follow the scene.

FramesIN → OUT04 / 07

From one sentence to a set of references.

One endpoint, three starting points. The number of images decides what they mean, so there's no extra mode flag.

0 images

Text to video

Just a prompt: framing, camera moves and mood all come from your words.

1–2 images

First / last frame

The first image opens the shot, the second closes it, Veo fills the motion in between.

≤3 images · fast-ref

Multi-reference

One reference each for character, style and set, so the same face holds across shots.

APIPOST /v1/videos/generations05 / 07

Two requests, one 15-second cut.

Standard REST, async tasks and a single Bearer key. Generation and extension share one endpoint; only the model name changes.

Submit, get a task_id back instantly
Poll status: pending → processing
completed, with the video URL

Get an API key Full docs

curl https://api.apipod.ai/v1/videos/generations \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "veo3-1-fast",
    "prompt": "Rain-soaked cyberpunk megacity at night, a drone glides between neon towers, 35mm, volumetric haze",
    "aspect_ratio": "16:9",
    "resolution": "1080p"
  }'

# {"code":200,"message":"success","data":{"task_id":"7f3a2b9c-4e1d-4f8a-b6c2-91d0e5a7c21e","status":"pending"}}

FAQQ & A06 / 07

Questions, answered

What is Extend?

Extend continues a finished Veo 3.1 video, taking an 8-second source to 15 seconds. Call veo3-1-fast-extend or veo3-1-quality-extend and pass the source video's task_id as remix_task_id.

Which videos can be extended?

The source must be a completed, regular Veo 3.1 generation from your own account, created within the last 48 hours. The fast line (fast, fast-4k, fast-ref) uses veo3-1-fast-extend and the quality line (quality, quality-4k) uses veo3-1-quality-extend. Lite videos and already-extended videos can't be extended yet.

What does the raw parameter do?

By default raw=false returns the full 15-second video, source plus extension. Set it to true to receive only the newly generated segment, handy when you stitch clips in your own editing pipeline.

Which aspect ratios and resolutions are supported?

16:9 landscape and 9:16 portrait, at 720p, 1080p or 4K (or use the -4k models for fixed 4K output). Extensions keep the source's aspect ratio unless you set one.

Does every video include sound?

Yes. Veo 3.1 natively generates sound effects, ambience and dialogue in sync with the picture, and the MP4 ships with its audio track.

How do I choose between Quality, Fast and Lite?

Quality delivers the best image for final output, Fast balances speed and cost for most work, and Lite is the cheapest way to explore ideas in bulk. Live prices for every tier are in the model list above.

How many reference images can I upload?

Fast and Quality take up to 2 images, used as the first and last frame. veo3-1-fast-ref takes up to 3 images as character, style or scene references.

Can I use the generated content commercially?

Content generated via the API is owned by the account holder and may be used commercially. See the Terms of Service for details.

Fin00:00:15:0007 / 07

Your next shot starts at second eight.

Veo 3.1 and Extend are live on APIPod, ready for production traffic.

Start creating Talk to us

Veo 3.1

Available Models

Veo 3.1 Lite 4K

Veo 3.1 Lite

Veo 3.1 Quality 4K

Veo 3.1 Fast 4K

Veo 3.1 Fast Reference

Veo 3.1 Quality

Veo 3.1 Fast

Eight seconds is no longer the end.

Keep the camera rolling.

Shoot the source

Submit the extension

Collect the cut

Editor's notes

One family, every job covered.

Pictures that bring their own sound.

From one sentence to a set of references.

Text to video

First / last frame

Multi-reference

Two requests, one 15-second cut.

Questions, answered

Your next shot starts at second eight.

Eight seconds
is no longer
the end.