Gemini, GPT, Flux, Seedance. Ask by name.
47 image, video, music and voice engines on one account. Pick one on screen or ask your Claude through MCP, see the cost before it runs and pay in Synapses. No five subscriptions to test one idea.
Every engine with what it does, what it costs and what already came out of it.
12 engines on screen
- Video
Gemini Omni
GoogleText, image or video: up to 15 s with 6 references of 2 characters, edits and extends a video of yours, 360p to 4K, with sound.
up to 15s · up to 4K · takes references · extends the take
from ⚡ 294 per take
Video@borderlessSeedance
ByteDanceCinematic with native audio, 4 to 15s. Takes references and a motion video.
- 2.0
- 2.0 Fast
- 2.0 Mini
- 2.5
up to 30s · up to 1080p · native sound · takes references
from ⚡ 800 per take- Video@borderless
Kling 3.0 Pro
KuaishouTop-tier motion to bring an image to life. 3 to 15s, and it carries a person's movement.
up to 15s · native sound
from ⚡ 2,100 per take - Video@borderless
WAN
AlibabaAn image that talks and sings: native voice, lip-sync and sound.
- 2.5
- 3.0
up to 10s · up to 1080p · native sound
from ⚡ 1,500 per take - Video
Kling Motion
KuaishouMoves your image with the motion of a video. Needs a person (torso visible) in both.
up to 10s · native sound · needs a driving video
from ⚡ 3,000 per take - Video
Shot Mimic
Recreates the reference video's shot with your character: same camera, same cuts.
up to 15s · up to 1080p · native sound · needs a driving video
from ⚡ 3,600 per take
Video@borderlessHailuo
MiniMaxMiniMax motion and physics at 768p, 6 or 10s. No audio, no references.
- 2.3
- 2.3 Pro
up to 10s
from ⚡ 2,100 per take
Video@borderlessMiniMax H3
MiniMax2K or 768p with native audio, 5 to 15s. More pixels per Synapse than the rest of the shelf.
up to 15s · up to 2K · native sound · takes references
from ⚡ 3,000 per take- Video
P-Video-2 Pro
PrunaMiniMax H3 accelerated by Pruna: up to 9 images, 3 clips and 3 audio references, and the take back in seconds, at 480p or 768p.
up to 15s · up to 768p · native sound · takes references
from ⚡ 660 per take - Video
FLUX 3 Max
Text or image, 5 to 20s at 720p or 1080p, with native audio and an end frame. Follows the prompt closely and comes back fast.
up to 20s · up to 1080p · native sound
from ⚡ 3,600 per take - Video
LTX Audio-to-Video
LightricksYour audio becomes the take: the track drives the lips and the gesture, and the photo opens the scene. Up to 20s, at 720p or 1080p.
- 2.5 Fast
- 2.5 Pro
- 2.3 Pro
up to 20s · up to 1080p · native sound · your audio becomes the take
from ⚡ 375 per take - Video
LTX
LightricksText or image, 6 to 20s at 720p or 1080p, with native sound and an end frame.
- 2.5 Fast
- 2.5 Pro
up to 20s · up to 1080p · native sound
from ⚡ 3,720 per take - +3moreMore engines for members.Sign in to see them
Two ways to ask. The same engine.
The App, with the engine on the shelf
Write the prompt, hang references from your gallery or your characters, pick the engine and see the cost before it runs. The piece lands in your archive with the recipe attached.
See the platform inside →Claude asks, the house generates
With the MCP on, your Claude (or Cursor, or Gemini CLI) sees the whole catalog, quotes the cost first and generates on your account. Paste this and see:
Use Sapiens to list the video engines that take a character reference, with the price of each. Then generate a 5-second take with the cheapest one, from the image I'm about to send. Before generating, tell me how many Synapses it will cost.
If it resonated,
we talk.
Screening by resonance. A real conversation, entry by curation. And what you build inside comes out published, with an address of its own.
Prefer Telegram? Talk there →Human screening within 48h. Direct communication, one on one.






