what AI model comes even close to Seedance 2.5 for realism right now?
because i still can’t believe this video cost 6 cents per second to generate. a 15-second version is 90 cents.
and you genuinely can’t tell it’s AI anymore.
if you’ve found something better, drop it below.
i don’t think most people can tell anymore whether a UGC video is AI-generated.
right now Seedance 2.5 is the best model i’ve used for AI UGC. the lip sync, the small pauses, the way the hands move: the usual tells just aren’t there.
it runs 6 cents per second of video. a 30-second ad works out to $1.80 to generate.
and making one is simple. it needs two things:
→ a reference video to take inspiration from
→ a good starting frame
get those right and the prompt barely matters.
recreated this AI UGC ad with Seedance 2.5 at 6 cents per second of video.
a 15-second ad at that rate costs 90 cents to generate. a full 30 seconds, $1.80.
and it came out ready to post. no editing pass, no reshoots.
that matters most when you’re testing variants. if every take is usable straight out of the model, ten 15-second hooks come to about $9.
watch the lip sync and the hands. that’s usually where AI UGC gives itself away.
made this ad in 10 MINUTES with Claude.
i gave it three things: an avatar, a script, and a creative direction.
it ran everything through an API i’d connected and picked the cheapest model for each step.
the video ran on Seedance at 9 cents a second
just told codex to set up a discord server for my new software company...
it asked me a few questions & in less than 10 minutes i had a complete server, built entirely by the AI
still insane to me that you can hand off something like this & have it done without hiring anyone or asking an expert
We’ve raised $1.3M from YC & Afore Capital so you never have to type or speak again.
Typing is slow. Talking into your computer is awkward. Subvocal fixes both
Our glasses give you the speed of speech without you making a sound or touching a keyboard.
It tracks your internal speech muscles with non-contact sensing and decodes silent, almost invisible mouthing into text at more than 200 words/min.
Pre-orders are now live at:
https://t.co/DzrwIpg3GR