Text to Video AI
Describe a scene in words and an AI model generates the whole clip: visuals, motion, camera work, and on several models native audio too. Arteza hosts Veo 3, Sora 2, Kling 3.0 Pro and Seedance 2.0 side by side, so one prompt can be tested across the strongest engines instead of committing to a single platform.
Made from a written prompt, nothing else
No source footage and no source image. Hover or tap to play, then try the prompt yourself.
MiniMax H3 Max 4cA polar research icebreaker cuts a slow furrow through pack ice in the Ross Sea at Antarctic midnight, the sky above erupting in curtains of electric-green and magenta aurora australis; wide locked shot from the bow watches the hull split floes apart, ice chunks rolling and glinting emerald in the auroral light, camera tilts up to let the aurora fill the entire frame, epic polar expedition cinematography.
Gemini Omni Flash 1.1 2cA master perfumer in a sunlit Grasse laboratory lifts tiny glass vials from a towering organ of scent bottles, uncapping each one to a bloom of invisible fragrance; slow lateral tracking shot glides across hundreds of amber bottles, camera pushes into an extreme close-up of a single drop falling from pipette into crystal flask, warm gold and pale ivory palette, artisan documentary cinematography.
Runway Gen-4.5 9cDive tender managing umbilical hoses on the back deck of a dive support vessel in a choppy swell, tracking shot moves alongside at deck level, hoses coil and tighten as the vessel rolls, sea spray bursts over the stern rail and catches orange safety-suit colour, the deck tilts rhythmically, horizon rises and falls behind, overcast North Sea light, gritty documentary photoreal
Kling O3 7cSkeins of freshly dyed wool lifted dripping from an indigo vat, the yarn oxidising from bright green to deep blue in the open air as it rises. Camera holds a close static frame at vat level, the dripping skein filling the upper half while indigo runs back into the dark surface below. The colour change sweeps across the wool continuously through the shot. Photoreal, saturated indigo against wet stone grey.
Wan 2.7 9cA single-engine crop duster banks hard above a Louisiana sugarcane field, chemical mist trailing behind it in a wide dissipating arc that slowly drifts and thins in the humid morning air. Locked off static wide shot, the plane entering frame from the left, pulling the mist arc across the green field below, the trail hanging then gradually settling downward into the crop rows as the plane exits frame right.
Kling 3.0 Pro 9cMing dynasty porcelain painters decorating unfired blue-and-white vessels in a vast kiln workshop, brushes tracing fluid cobalt dragons and peonies onto pale clay as pigment bleeds softly at each stroke, camera tilting slowly upward from hands and turning vessels at bench level past rows of apprentices and drying racks to towering kiln arch windows flooding the hall with cool diffused daylight
How it works
Write the shot
Good video prompts cover four things: the subject, the action, the setting and the camera. Describing a handheld camera following a cyclist through a rainy street gives the model far more to work with than just naming a cyclist.
Pick a model and settings
Choose an engine, duration and aspect ratio. Each model card shows its live credit cost, so you know exactly what a clip will cost before you commit.
Generate, compare, cut
Clips arrive in seconds to a couple of minutes depending on the model. Rerun the same prompt on a second engine to compare takes, then sequence the winners into a longer edit.
Models you can use right now
Every model below is live on Arteza with its current credit cost, pulled from the same pricing engine the studio uses at generation time.
Seedance 2.0
Cinema-grade AI video with native audio synthesis
from 19 credits
Frequently asked questions
Which text to video model should I start with?
Veo 3 if the scene needs spoken dialogue and 1080p delivery, Sora 2 for longer takes with believable physics, Kling 3.0 Pro for polished cinematic clips with native audio at a lower cost, and Seedance 2.0 when you need fine control over framing and aspect ratio. All four run on the same Arteza credit balance.
How long can the videos be?
It varies by model: most engines generate roughly 4 to 10 seconds per run, and Sora 2 goes up to 20 seconds. For longer pieces, creators generate multiple shots and cut them together, or extend a clip that is almost long enough.
Do the videos include sound?
On some models, yes. Veo 3, Kling 3.0 Pro and Seedance 2.0 generate native audio with the clip, and Veo 3 additionally targets synchronized dialogue. Other models output video only, which suits workflows that add music or voiceover in the edit.
How much does text to video cost?
Each model has its own per-clip or per-second credit cost, shown live on this page and in the studio before you generate. You start with free credits on signup, so you can test prompts before spending anything.