Private AI image generation — prompts and outputs that stay yours
Most AI image tools publish your work to a community feed or wrap every prompt in a review layer that second-guesses your intent. LivePair's private studio works differently — this is what 'private' means here, concretely.
No public feed
Generations in the private studio are yours — they are not indexed, not shown in a community gallery, and not used to decorate the homepage. The public gallery at /hub is a separate, explicitly curated surface fed by the site's own showcase pipeline, not by user renders.
If you want a model's output to exist only in your account, that's the default here — not a settings toggle.
Clear rules, stated once
LivePair's own prompt layer blocks only the categories that are illegal everywhere — content involving minors and sexual violence. Everything else goes to the model, and the model's provider applies its own alignment — which for the private model line is deliberately minimal.
One more real limit worth knowing up front: uploads of photos depicting real people are rejected by an image check — the tool works on text prompts and its own generated images only.
The model line
The studio runs on hosted image and video models — about 20 in the current catalog — including a private image-to-video line that mainstream apps don't carry at all. Model choice is explicit: you pick the model, see its price per result before you run it, and get billed per result, not per subscription.
Two ways to pay: prepaid credits on an account, or per-call USDC through the x402 endpoint without ever making an account — the API path is documented at /docs.
What it is not
It's not a free tool — generation is paid per result because private compute costs real money. It's not 'anything goes' — the illegal categories are enforced at the platform layer, not left to the model. And it's not a filter lottery: the rules above are the whole rules — there's no hidden second moderation layer guessing at your intent.