Questions people actually ask
Accounts, what the free tier includes, how long anything is kept, what appears on your output, and where to change a plan. Answered at the length the answer needs rather than the length that sounds reassuring.
Getting an account
- Do I need an account to look around?
No. The home page, the guides, the safety page and the tool pages are all readable signed out, and the demo on the home page was made with this product.
You need an account to run a job, because an anonymous generate button is a free compute faucet — one forum link away from being somebody else's render farm.
- What do you ask for at signup?
A username, an email address, a password, and your date of birth. No real name, no address, no phone number.
- Why the date of birth?
The service is for adults and registration refuses anyone declaring an age under 18. It is a self-declaration rather than verification — we would rather say that plainly than imply a check we do not run.
- Can I delete my account?
Yes, from your profile. Deletion erases your rows across every table that holds them, and there is an automated test that fails the build if a new table holding user data is added without being covered by it.
The one exception is the incident record described below, which is kept when a request crosses one of the four hard limits.
The free tier
- Is the free tier a trial?
No. It is a real plan with no expiry: the same tools, the same models and the same rules as a paid plan, with lower limits.
- What does paying actually buy?
Compute. A subscription raises your allowance — it does not unlock a different product, a different model set or a different answer to what we refuse to generate.
- How is the allowance measured?
As a rolling weekly budget of GPU time rather than a monthly one, so a heavy week does not shut the door on you for the next twenty-eight days.
Images, video and chat draw on the same budget. A quiet week of pictures means more conversation, and the reverse.
- Is chat included on the free tier?
Yes, on the same allowance as the picture tools. Your account page estimates how many more replies are in this week's budget, based on your own recent conversations rather than an average nobody matches.
- Where are the limits written down?
Your own current plan and its allowance are on your profile, which is the figure that is actually being applied. Plans and prices are on the pricing page. What is refused, on every plan, is on the safety page.
What happens to your files
The short version: this is a workshop, not an archive, and that is deliberate. The privacy policy has the full detail and names the mechanism behind each line.
- How long do you keep what I make?
Uploaded photographs are deleted when the job finishes — overwritten with seven random passes and a zero pass rather than merely unlinked — and a janitor removes anything missed after two hours.
Generated files are erased when you delete them and swept automatically after twenty-four hours. Prompt text is blanked on the same schedule, so the job record survives for your gallery but the words do not.
- So I should download things?
Yes. Treat every result as something to save, not something stored for you. Nothing here is backed up on your behalf.
- Is my work sent to OpenAI, or Stability, or anyone else?
No. Prompts, uploads and conversations are processed only on infrastructure we operate, control and administer, and are never handed to a third-party AI provider.
The single exception is a web search in chat, which you turn on per message and which by definition reaches the public internet. You are told at the moment you turn it on, and the reply cites what it found.
- Do you train on my work?
No. Not your prompts, not your uploads, not your conversations, not your outputs.
- Is anything kept on purpose?
One thing. When a request crosses one of the four hard limits we record the prompt text, the terms that matched, the tool used and the client IP address, and that record is not swept on the ordinary schedule.
It is encrypted at rest and readable only by an administrator. A safety control that forgets is not a safety control, and a privacy policy that omits the one category a service knowingly keeps is not credible about anything else on the page — so it is written out in full in ours.
- Do you count visitors?
Yes, anonymously. What is stored is one number per visitor per day, derived from a salted digest rather than an address, with the salt regenerated daily and never written down. It can answer how many people came; it is not capable of answering who, or what they did.
What it will and will not make
- Is there adult content?
No, and there is nothing to unlock. Sexual content and nudity are absent from the software we run rather than switched off in it — no setting, plan, administrator action or support request introduces them, on the site or on the API.
That is a business decision rather than a moral one. It is what lets an ordinary payment processor work with us, and it is why the rest of the product can be as permissive as it is.
- What is actually refused, then?
Four things, each traceable to a law rather than to our taste: sexual content involving minors; sexual imagery of a real, identifiable person made without their consent; content made to deceive about a real person in a way that causes harm; and material that would give meaningful assistance toward mass casualties or serious violent crime.
The full list, with the statute behind each item, is on the safety page. It is four items long and it does not grow quietly.
- Is political content blocked?
No. Political subjects are permitted by default and are not filtered by viewpoint — satire, protest imagery, campaign parody and criticism of any government, party, movement or figure, in every direction.
That is tested rather than promised: the build pairs mirrored prompts from opposing sides of political, religious and identity debates and fails unless each pair is treated identically.
An optional Strict profile, meant for classrooms and shared machines, does exclude political subjects along with violence and other mature themes. It is off unless you turn it on.
- Are the chat models neutral?
Not entirely, and we would rather name it. They are third-party open-weight models and carry whatever biases their training gave them. We add no political layer of our own and our own refusal logic is symmetric, but we cannot remove a lean baked into a model's weights — the current models are known to push back on criticism of the Chinese government.
- Something was refused and I think that is wrong. What now?
Use the appeal page if the refusal is attached to your account, or the contact page otherwise. Refusals outside the four fixed limits are usually a profile setting on your account, which you can change yourself.
Your output
- Is there a watermark?
No. Your output carries no branding of ours.
The one mark that ever appears is a small AI-generated label on an image that depicts a real, identifiable person. That is disclosure — a viewer is entitled to know an image was generated — and it is what the relevant statutes ask for.
- Who owns what I make?
As between you and us, you do. We claim no ownership of your output and we do not use it to train anything.
We cannot promise a result is free of third-party rights: generative models can reproduce protected elements, and whether a given picture is yours to use commercially depends on facts we cannot see. That judgement is yours.
- Can I use it commercially?
Nothing in the plan stops you. See the answer above about third-party rights, which is the real constraint and is not one we can resolve for you.
API and your own tools
- Is there an API?
Yes, on the same account. The chat surface is OpenAI-compatible, so an editor plugin, a script or a terminal client works with a settings change and no plugin from us. There is a ComfyUI-protocol surface for image work as well.
The API docs are at /docs, and the same limits apply there as in the browser.
- Do the rules change on the API?
No. The four fixed limits are compiled into the software, and adult content is absent from it. There is no API-only mode with a different answer.
Billing
- Where do I subscribe, change tier or cancel?
All three are on your profile. It also shows what your own plan costs — that figure is the one being charged, so if it ever disagrees with the pricing page, believe your profile.
- Do you store my card?
No. Card details go directly to Stripe, our payment processor, and never touch our servers. We hold a Stripe customer reference, the subscription state and your plan tier.
- What happens when I cancel?
The subscription ends at the close of the current paid period. You keep paid entitlements until then and the account returns to the free plan afterwards. Cancelling a plan does not delete your account.
- Are prices inclusive of tax?
Prices are shown in US dollars. Sales tax, VAT or similar charges are added where we are required to collect them, and the amount due is shown before you confirm payment.
Where the authoritative version lives
This page is a plain-language summary and it is meant to be accurate, but it is not the governing text. What we refuse is set out with the law behind each item on the safety page; what is kept and for how long is in the privacy policy; the agreement itself is in the terms. Your own plan and its allowance are on your profile. If any of those disagrees with this page, they are right and this page is a bug worth telling us about through the contact page.