LoRA Rebate Upgrade: No More Cap, Keep Earning Over Time
Train and publish a LoRA on PixAI, and every time another creator uses it to generate images, you earn a rebate in PixAI credits, paid out as a share of what they spend. We’ve now overhauled the whole rebate system. It gives you more room to earn, and it’s built for steady, long-term growth.
— POINT ONE —
Table of Contents
What This Upgrade Changes
▸ THE BIGGEST CHANGE
We’ve removed the rebate cap
Under the old system, each LoRA could earn a lifetime maximum of 200,000 credits in rebates. Once it hit that cap, it stopped earning, no matter how many more times it was used.
That cap is now gone. In return, the rebate rate now steps down gradually as your cumulative rebate base grows (starting at 5% and tapering as usage increases). But with no overall ceiling, your LoRA keeps earning rebate for as long as other creators keep using it. Over time, the total you can earn actually ends up higher than before.
One important point: this upgrade only adds to your earnings, never reduces them. The most you could earn under the old system (equivalent to the 200,000-credit cap) is completely unaffected—you’ll still receive it in full. The tapering rate applies only to the portion beyond the old cap, which you couldn’t earn at all before. There’s nothing for you to calculate—for creators, this is a pure gain.
— POINT TWO —
Who Earns a Rebate
A single generation earns the LoRA’s creator a rebate when all of these are true:
CONDITION 01
The LoRA was trained on PixAI (models imported from elsewhere don’t count).
CONDITION 02
It’s used by someone other than the creator (using your own LoRA earns no rebate).
CONDITION 03
The task completes normally and actually spends credits.
In short: the more other creators love and use your LoRA, the more you earn.
— POINT THREE —
How the Rebate Is Calculated
Each week, we add up how much each of your LoRAs was used and calculate your rebate from that. A few factors work together to determine the amount:
FACTOR 01 · USAGE
Total usage
When other users generate images with your LoRA, the credits those tasks spend form the basis for your rebate (we call this your rebate base). The more your LoRA is used and the more credits those tasks consume, the larger your rebate base.
FACTOR 02 · TIERED RATE
Tiered rebate rate
Rebates use a tiered rate: as a LoRA’s cumulative rebate base (the credits spent using it) grows, each additional portion is paid at a progressively lower rate.
FACTOR 03 · MULTI-LoRA
Number of LoRAs used in one image
When a single image uses a lot of LoRAs at once, we adjust the rebate rate slightly. The point is to stop people from stacking large numbers of LoRAs just to inflate their rebates, not to chip away at what regular creators earn. When we make this adjustment, we still try to protect each creator’s fair share, and normal combinations of a few LoRAs are barely affected.
FACTOR 04 · PAYING USERS
Real paying users
Once a LoRA’s cumulative rebate base reaches a higher tier, it needs to reach a minimum number of real paying users before it earns the full rebate at that tier. This steers rewards toward the LoRAs people genuinely love, and helps prevent abuse such as artificially inflated usage.
Worth noting: within the range the old system already covered—that is, before your cumulative rebate reaches the old cap of 200,000 credits—no paying users are required. In other words, your baseline earnings under the new system are exactly the same as before, and you won’t earn less than you used to just because a paying-user condition was added.
📌 NOTE
The amount actually credited factors in everything above, so it isn’t as simple as usage × 5%. The rates in the table are the starting point for the calculation, and your final payout is determined at weekly settlement.
— POINT FOUR —
Why These Rules Exist
Removing the cap was a promise we made to all creators at our third anniversary, and we’ve always wanted to deliver this rebate to everyone who creates with care—steadily, and for the long term.
But protecting that intention isn’t easy. The richer the rebate, the more attempts there are to exploit it for unfair gain. All along, a small number of people have profited not through their work, but by gaming the system—creating large numbers of fake accounts, stacking LoRAs for no real purpose, or even deleting a LoRA right after collecting its rebate and republishing it to farm rebates over and over. Every credit gained this way quietly dilutes the rewards that genuinely dedicated creators have earned, and we’ve put significant time and effort into preventing it and returning those resources to the people who deserve them.
So the rules you see in the new system exist to keep the rebate where it belongs: with creators who put real care into their work and value their own creations and fans. That bond between a creator and their fans is exactly what we most want to protect.
— POINT FIVE —
When Rebates Are Paid Out
Rebates are settled once a week (based on Japan Standard Time, JST).
▸ Each settlement cycle is one full week, from this Monday to the next.
▸ The rebate earned during that week is paid into your account early the following Monday.
▸ Even if your LoRA only starts getting used partway through the week, those days still count toward that week and are paid out on the Monday along with the rest. Nothing rolls over to the next week.
▸ The exact payout time may vary slightly depending on system conditions, so please go by what actually arrives in your account.
💡 You can review your rebate details under Profile → Membership & Credits → Credit History by selecting “LoRA Usage Rebate”.
— POINT SIX —
About Joining the Rebate and Keeping Your LoRA
To keep rebate records intact and prevent the abuse described above, we’ve made the following changes to the “join rebate” setting for LoRAs:
▸ From now on, newly published LoRAs are not enrolled in the rebate by default; if you’d like yours to take part, you’ll need to turn it on manually when publishing.
▸ Once a LoRA is enrolled, it can no longer be deleted. This prevents behavior like “collect the rebate, then delete,” which harms both the platform and its users.
▸ Existing LoRAs are enrolled in the rebate by default, with a 2-week grace period during which you can freely choose whether to keep them enrolled.
▸ After the grace period ends, an enrolled LoRA’s status is locked—it can no longer be withdrawn from the rebate, nor deleted.
💡 If you’d like to keep the option to delete a particular LoRA, please adjust its setting within this 2-week grace period.
— TIPS —
How to Keep Earning Rebates
Polish your LoRA
Make it genuinely useful with a recognizable style — the kind creators come back to again and again. The more it’s used, the larger your rebate base.
Keep an eye on new architectures
When you retrain a LoRA for a new architecture, it starts over as a new version and begins accumulating its rebate again, so it qualifies for the higher rate once more. Keeping up puts your LoRA in front of more users and helps you earn more over the long run.
— FAQ —
Frequently Asked Questions
What happens to the rebates my LoRA had already accumulated before the upgrade?
The rebates you’ve already earned won’t be reset to zero. After the upgrade, the system automatically carries over your LoRA’s existing rebate total, so it keeps building under the new rules instead of starting from scratch. If your LoRA supports several model architectures, the system follows the priority order SDXL → DiT.1 → DiT.2 → SD1.5 and picks the highest one on that list to inherit this historical total; the remaining architectures start fresh. This is designed to encourage everyone to train LoRAs for new architectures, and to reward creators who have already adapted theirs to several.
Is the rebate rate always 5%?
Not necessarily. 5% is the top tier. Your actual rebate steps down as your LoRA’s cumulative rebate base grows, and it’s also affected by factors like how many LoRAs were used in the same image, so the final amount is determined at weekly settlement.
My LoRA used to be at 5%. Why is the rate lower now?
The rate is tiered: as your LoRA’s cumulative rebate base climbs past certain thresholds, the new portion is calculated at the next rate down (for example, from 5% to 4%). This doesn’t touch the rebates you’ve already received, and it doesn’t mean you’re earning less. With the overall cap now gone, your long-term total keeps growing as long as your LoRA stays in use.
Why is the rebate sometimes different even when usage is the same?
The rebate factors in several things, including how many LoRAs were used together in a given image. If your LoRA is stacked alongside many others, the rebate from that use is split among them, so your share is smaller.
Who counts as a “paying user”?
Anyone who has made a purchase on PixAI (with the exception of the $1 credit pack). What matters is whether the user is a paying user, not whether they happened to use free or paid credits when using your LoRA.
How do paying users affect my rebate?
Once a LoRA’s cumulative rebate base reaches a higher tier, it needs to reach a minimum number of real paying users before it earns the full rebate at that tier. In the 5% tier, which the old system already covered, no paying users are required. In other words, your baseline earnings under the new system match the old one exactly, and you won’t earn less than before just because a paying-user condition was added. The point is to steer rewards toward the LoRAs people genuinely love and to prevent abuse such as artificially inflated usage. You don’t need to track any of this yourself; the system counts it automatically at settlement.
Do I earn a rebate when I use my own LoRA?
No. A rebate is only generated when other users use your LoRA.
Do images generated with free credits earn the creator a rebate?
Yes. As long as the use meets the rebate conditions, it counts toward the creator’s rebate whether free or paid credits were used.
I published my LoRA partway through the week. Do the remaining days count?
Yes. Rebates are settled by calendar week. Your LoRA starts counting toward the week the moment it’s first used, and it’s settled and paid the following Monday along with everything else.
When does the rebate land in my account?
Once a week. Each settlement cycle runs from one Monday to the next, and that week’s rebate is paid into your account early the following Monday. The exact timing may vary slightly depending on system conditions.
What’s the benefit of retraining a LoRA for a new model architecture?
When you retrain the same LoRA for a new architecture, it starts over as a new version and begins accumulating its rebate again, so it qualifies for the higher rebate rate once more. Keeping up with new architectures helps you earn more over the long run.
⚠ IMPORTANT NOTES
To keep the rebate system healthy and sustainable, we may adjust the rebate rates and related rules from time to time. Should we identify any abuse of the system, we also reserve the right to introduce additional restrictions.
— START EARNING —
Train a great LoRA and let it keep earning for you
The cap is gone, so a great LoRA’s earnings are no longer ceilinged. Train your LoRA on PixAI now, or start generating your next piece.
