- Per clip
- 87 cr / 6s
- Frame
- 1080p
- Max length
- model's own
Models
One Render card, every model.
Pick the renderer per approve. The studio writes the scene once, draws the keyframe once, and sends it to whichever model you chose — with the price on the card before you press it.
Standard tier
On every plan. Fast, cheap, and where most scenes should start.
Unlocked by Starter
- Per clip
- 130 cr / 6s
- Frame
- 720p
- Max length
- model's own
- Per clip
- 173 cr / 6s
- Frame
- 720p
- Max length
- model's own
- Per clip
- 120 cr / 5s
- Frame
- 720p
- Max length
- model's own
- Per clip
- 168 cr / 5s
- Frame
- 1080p
- Max length
- model's own
Creator tier
Creator and up. Stronger motion and people, audio on Seedance.
Unlocked by Creator
- Per clip
- 291 cr / 5s
- Frame
- 720p
- Max length
- model's own
- Per clip
- 365 cr / 5s
- Frame
- 720p
- Max length
- model's own
- Per clip
- 568 cr / 5s
- Frame
- 720p
- Max length
- model's own
Image models
One composer, every image model.
Switch the composer to Image, pick the model in the picker with its credits per still beside it, and the still lands on your Assets wall. Each model has its own page.
LTX 2.5
Picture and sound in one pass. Draft on LTX 2.5 Fast, up to 20 seconds and 4K, then finish on Pro, with every shot held to your reference photos.
Wan 3.0
Motion that holds together: cloth that spins, water that splashes, people who move like people. Any length from 2 to 10 seconds a shot, from your own reference photos.
Kling 3 Turbo Pro
People who move like people, at 1080p: faces that turn and laugh, hands that work, bodies that dance. Any length from 3 to 15 seconds a shot, vertical for the feed.
Seedance 2.5
Write a scene, anchor every shot on your reference photos, and render it on Seedance 2.5, 4 to 30 seconds a clip with sound, up to 1080p.
Veo 3.1
Google's model for scenes you hear as well as see: a spoken line on the right frame, the rain, the room. Turn the sound on.
Nano Banana Pro
Gemini's image model; strong on references and text.
FLUX.2 Pro
FLUX's newest; takes reference images.
Seedream 4.5
ByteDance; photographic, about 4 megapixels, generates and edits with up to 10 references.
GPT Image 2.5 Flare
OpenAI's fast default; text, layouts and edits, up to 16 references.
Ideogram 4.5
Graphic, typographic, poster-like. Text-only.
Keyframes
Every clip anchors on a still you approved.
Picking a scene draws its first frame from the references you attached. That still is what the clip renders from, on every model — so the face, the room and the wardrobe are decided before a credit is spent, not guessed by the video model.
One renderer
Every model, one bill in credits.
Kling, LTX, Wan, Seedance and Veo all render through fal, so there is nothing to connect and no second bill. The Render card shows each clip's price in credits before you approve it.
See the plans