PIXQUILL
Free credits on the house

10 credits on sign up, 5 credits daily.

NEW

Flux 3 Klein and Kling 4.0 are live. Free credits to start.

✦ VIDEO AGENT

YOUR AI VIDEO AGENT FOR EVERY STORY

Start with an idea or a still. Describe your scene and make a video that looks and moves the way you pictured it.

10+ MODELSVIDEOS UP TO 30SUP TO 4K

PLANS FROM $4.99/MOPACKS FROM $9.99SEE PRICING →

NOW SHOWING
VIDEO
BLADE MIST
TRY IT NOW

Start Creating

0 / 2000

COST—
▸ PRICED

8S VIDEOS FROM 60 CREDITSPLANS FROM $4.99/MOPACKS FROM $9.99

SEE PRICING →
INPUT
A PROMPT, FRAMES, OR REFERENCES
OUTPUT
VIDEOS UP TO 4K, MOST WITH SOUND
A RUN TAKES
1 TO 3 MINUTES
THE ENGINES
VEO, SEEDANCE, KLING, MORE

PLANS & PACKS

UP-FRONT PRICING

Recommended packs start at $9.99 and stay valid while your account is active. Memberships start at $4.99 a month. One balance covers every image and video tool. The console prices every video before you render.

BEST VALUE

Standard Pack

$9.99

700CREDITS

≈ 350images

≈ 30SOTA images

≈ 88svideo

≈ 43sSOTA video

A full issue's worth of art, plus videos for the launch.

ONE PAYMENT · NEVER EXPIRE

Pro Pack

$19.99

1,600CREDITS

≈ 800images

≈ 69SOTA images

≈ 208svideo

≈ 100sSOTA video

Batch season. Whole chapters and the videos that sell them.

ONE PAYMENT · NEVER EXPIRE

MOST POPULAR

STANDARD MEMBERSHIP

$4.99/ month

1,000CREDITS EVERY MONTH

For weekly creators and side projects.

≈ 500images

≈ 43SOTA images

≈ 128svideo

≈ 62sSOTA video

NO LOCK-IN · CANCEL ANY TIME

5× STANDARD

PRO MEMBERSHIP

$19.99/ month

5,000CREDITS EVERY MONTH

For daily image and video production.

≈ 2,500images

≈ 217SOTA images

≈ 664svideo

≈ 312sSOTA video

NO LOCK-IN · CANCEL ANY TIME

SEE ALL PLANS →

Membership credits are available while the membership is active. Unused membership credits expire when the membership ends.

Image estimates use Flux 3 Klein at 1K, the lowest-priced run on the site. Video estimates use 8-second Veo 3.1 Lite videos at 720p, or Wan 3.0 at 480p for the SOTA row. The console shows the exact cost before every run.


SHOWCASE

FRESH OFF THE BENCH

Real output from the engines in this console. Videos play as you scroll, tap any card for the full cut.


USE CASES

WHERE THE VIDEO GOES

Six jobs the console handles daily, one cover frame each. Match yours, describe the shot, hit generate.

Black knit sneakers stacked in drifting fog against deep jungle green, lit like a studio product shot

PRODUCT VIDEO

A packshot becomes a looping hero shot. Put the product in drifting fog, on open water, under hard studio light, then orbit it or push in.

Rainy city street at night with neon signs reflecting on wet asphalt and pedestrians under umbrellas

SOCIAL VIDEOS

Reels and Shorts run on mood. Render vertical 9:16 straight from the console, one hook per video, from rainy neon streets to slow morning light.

Two paper coffee cups on the polished hood of a vintage muscle car, one tipped over with coffee pooling

AD CONCEPTS

Test directions before anyone books a shoot. Generate the coffee-on-the-car-hood frame and nine other takes, keep the two that survive review.

Soldiers marching through heavy fog across an open field, the lead soldier raising his arm to signal

STORYBOARD PREVIZ

Lock the first and last frame and the model fills the motion between. Previz a march through fog or a full action beat before the shoot day.

Close portrait of a woman with a wet-look blonde bob lit by red neon rim light in a dark studio

CHARACTER WORK

Reference images hold the same face across shots. Take a character from still portrait to moving performance, lit any way the scene needs.

A veiled figure in an ornate golden crown, gold leaf details catching low-key light against black

MUSIC VISUALS

Give a track its look: veiled figures, gold light, slow reveals. On models that take audio reference, the cut follows the music.


THE STUDIO

EVERY TOOL, ONE BALANCE

A video usually starts or ends as an image. The rest of the studio shares your uploads and credits.


HOW IT WORKS

MODE, SHOT, VIDEO

    • TEXT TO VIDEO
    • FIRST FRAME
    • START + END FRAMES
    • MULTI-IMAGE REF
    • OMNI REF

    PICK A MODE

    Text to video, first frame, start and end frames, multi-image reference or omni reference. The console only asks for the attachments that mode takes.

  1. A cyclist descends a mountain road at dawn, slow drone follow

    MODELMINIMAX H3DURATION5SRATIO16:9RESOLUTION768P

    WRITE THE SHOT

    Say the subject, the motion and the camera in plain language, and @mention your uploads when it helps. Pick a model, a duration, a ratio and a resolution: the exact credit price shows before you generate.

  2. Film still from a generated video: a cyclist descending a mountain road at dawn

    video-cyclist-dawn.mp4✓ 5S · 768P · READY

    GET THE VIDEO

    A run usually takes 1 to 3 minutes. The finished video lands in the console and in your library, ready to download.


THE ENGINES

PICK YOUR ENGINE

10+ video models share one console and one credit balance. The picker pins the featured six on top and groups the rest by vendor. Specs and the entry price sit on each card; every run is priced before you commit.

DEFAULT

MINIMAX H3

The model the console opens on. Sound comes with every video, up to 5 reference images are supported, and per-second billing keeps short drafts cheap.

768P-2K · AUDIO ON · 4-15S · 5 REF IMAGES

FROM 16 CREDITSPER SECOND

NEWEST KLING

KLING 4.0

Kuaishou's newest Kling. Up to ten reference images or a short clip hold one subject or a style across a run, and first and last frames lock in the move between them. Audio comes with every video, runs reach 30 seconds, and output tops out at 1080p.

3-30S · ALL 5 MODES · 720P-1080P · AUDIO ON · 10 REF IMAGES

FROM 57 CREDITSPER SECOND

SYNCED DIALOGUE

VIDU Q4

ShengShu's cinematic model is built around references: a first frame or up to 15 images, plus up to three voice tracks. Dialogue and sound arrive in the same pass as the picture, already in sync. Pick it for character scenes that need to speak.

540P-4K · AUDIO OPTIONAL · 3-16S · 15 REF IMAGES

FROM 15 CREDITSPER SECOND

BUDGET VEO

VEO 3.1 LITE

The entry tier of Google's Veo 3.1 line. Audio comes standard, output reaches 4K, and pricing is per video, so an 8-second cut costs the same as a 4-second one at the same resolution.

720P-4K · AUDIO ON · 4/6/8S · PER-VIDEO PRICE

FROM 60 CREDITSPER VIDEO

LONG RUNS

WAN 3.0

Alibaba's video flagship. It covers all five modes, runs from 2 to 30 seconds, and accepts images, videos or audio as references. Sound is a switch you control.

2-30S · ALL 5 MODES · 480P-1080P · 10 REF IMAGES

FROM 16 CREDITSPER SECOND

LATEST SEEDANCE

SEEDANCE 2.5

ByteDance's newest Seedance covers all five modes and takes the largest reference load on the bench: up to 30 images, or 10 videos and 10 audio tracks. Runs go from 4 to 30 seconds, and runs that use video references bill at a lower per-second rate.

4-30S · ALL 5 MODES · UP TO 30 REF IMAGES

FROM 56 CREDITSPER SECOND

4K FLAGSHIP

VEO 3.1 QUALITY

Google's highest-fidelity video model and the most expensive engine on this page. Sound is always on, output reaches 4K, and videos run 4, 6 or 8 seconds. Reserve it for the cut that has to hold up on a big screen.

720P-4K · AUDIO ON · 4/6/8S · PER-VIDEO PRICE

FROM 500 CREDITSPER VIDEO

OMNI REFERENCE

KLING 3.0

Kuaishou's flagship runs all five modes, including omni reference, where images, video and voice join one brief as elements. Videos run 3 to 15 seconds, output reaches 4K, and the audio track is optional.

3-15S · 720P-4K · 5 MODES · AUDIO OPTIONAL

FROM 28 CREDITSPER SECOND

MID TIER

VEO 3.1 FAST

The mid tier of the Veo 3.1 line. It carries the same audio and 4K ceiling as Quality, adds multi-image reference that Quality does not take, and prices per video between the two.

720P-4K · AUDIO ON · 4/6/8S · 3 REF IMAGES

FROM 120 CREDITSPER VIDEO

PREVIOUS GEN

SEEDANCE 2.0

The previous Seedance generation. It covers all five modes, reaches 4K, and runs 4 to 15 seconds at a lower per-second rate than 2.5.

4-15S · 480P-4K · ALL 5 MODES · 9 REF IMAGES

FROM 38 CREDITSPER SECOND

SILENT OUTPUT

HAPPYHORSE 1.1

Alibaba's motion model without an audio track: the video leaves sound to your edit. Three input modes, 720p and 1080p output, and up to 9 reference images. Made for B-roll you will score in the edit anyway.

3-15S · 720P/1080P · NO AUDIO · 9 REF IMAGES

FROM 45 CREDITSPER SECOND

DRAFT TIER

SEEDANCE 2.0 FAST

The draft tier of the Seedance line. Every mode is open at 480p and 720p, and video references bill at a further per-second discount. Iterate here, then render the keeper on a 4K engine above.

4-15S · 480P/720P · ALL 5 MODES · 9 REF IMAGES

FROM 24 CREDITSPER SECOND


TIPS

PROMPT FOR MOTION

  • 01Name the subject, the motion and the camera, in that order. 'A paper lantern rises over a night market, slow tilt up' beats a pile of adjectives.
  • 02One video, one beat. A run of a few seconds up to 30 holds a single action; asking for a plot gets you mush.
  • 03Say the camera move out loud: static shot, dolly in, orbit, handheld follow. Models take direction literally.
  • 04Lock the look with a first frame, then ask only for motion. Image to video is the cheapest way to keep a style.
  • 05For start and end frames, pick two images that could plausibly be neighbors. The model fills the gap, not miracles.
  • 06Draft at 480p or 720p on a value model, then render the keeper on a SOTA one. Rerolling at 4K is the expensive way to move one cloud.

ABOUT

WHAT THE VIDEO AGENT IS

An AI video generator turns a prompt, a frame or reference material into finished video, no shoot and no edit suite. PixQuill's Video Agent is the console we built around that idea: 10+ models on one bench, for people who need videos this week rather than a new hobby. Marketers test ad directions, sellers animate product pages, creators feed the feed.

Start a video the way the idea arrives: a sentence, one frame, two frames, a set of stills, or a mix of images, video and audio. Everything runs on our servers, so there is nothing to install, and the console prices the run before you commit. Durations run from a few seconds up to 30, with 4K at the top of the bench.

A run usually takes 1 to 3 minutes, longer renders finish in the background and land in your library, and a run that fails on our side refunds its credits automatically. Credits are shared with every other PixQuill tool, so the image you made in the Image Agent can be the first frame here.


FAQ

STRAIGHT ANSWERS

▸ What is an AI video generator?

A tool that turns a prompt, a frame or reference material into a finished video. PixQuill's Video Agent runs 10+ models behind one console: you write or attach the input, pick a model, and the video renders on our servers, usually in 1 to 3 minutes.

▸ Which video models does PixQuill run?

PixQuill runs 10+ video models: Google's Veo 3.1 in Quality, Fast and Lite tiers, ByteDance's Seedance 2.5, 2.0 and 2.0 Fast, Kuaishou's Kling 4.0 and Kling 3.0, Alibaba's Wan 3.0, HappyHorse 1.1 and MiniMax H3 (the Hailuo line), plus ShengShu's Vidu Q4 with synced dialogue. MiniMax H3 is the default; the picker puts featured picks on top and groups the rest by vendor.

▸ Text to video, image to video, reference to video: what is the difference?

Text to video starts from a prompt alone. First frame and start+end frames animate from one or two stills, which is what image to video means here. Multi-image reference holds a subject or style steady across several stills, and omni reference reads images, videos and audio together on the models that support it.

▸ How long are the videos, and how sharp?

Durations depend on the model, from a few seconds up to 30. Resolution runs up to 4K on Veo 3.1, Seedance 2.0 and Kling 3.0, with 720p and 1080p covering most of the rest. The console only offers what the picked model can actually do.

▸ Do the videos have sound?

On most models, yes. Veo 3.1, MiniMax H3 and Kling 4.0 always render with audio. Seedance, Wan 3.0 and Kling 3.0 take sound as a switch you control per run. HappyHorse 1.1 renders silent.

▸ How much does a video cost?

An 8-second Veo 3.1 Lite video at 720p costs 60 credits. Per-second models price by model, duration and resolution, with Wan 3.0 starting at 16 credits per second at 480p. Reference inputs and audio can also change the price. The console shows the exact cost before you generate. Packs and memberships are on the pricing page.

▸ How long does a run take, and what if it fails?

Usually 1 to 3 minutes. Longer renders keep going in the background and land in your library. If a run fails on our side, the credits refund automatically, in full.

▸ Can I close the tab while a video renders?

Yes. Rendering happens on our servers, so the run continues after you close the tab. When it is done, the video is in your library under Works. You can get there from the sidebar or your avatar menu.

▸ Can I use the videos commercially?

Videos you generate are yours to use, including in commercial projects, within our terms of service. Rights in anything you upload stay where they were, so only attach material you own or have permission to use. Details live on the Copyright page.


READY

GIVE THE IDEA MOTION

Write the shot or pin it with frames, pick a model, and watch the console price the run. New accounts start with free credits.