How usage works
No credits. No coins. No tricks. Your plan includes a guaranteed number of generations every month, and when the GPUs are quiet you can generate as much as you like on top.
The meter
Instead of buying tokens and watching a balance tick down, you get a meter: a percentage of your monthly allowance. Generating uses some of it, the amount depending on how heavy the model is and how busy the platform is right now. It resets every billing cycle.
Every plan advertises a guaranteed minimum. That is the worst case: the number of generations you can run on the heaviest model your plan allows, at the busiest time of day. Almost everyone lands well above it, because a lot of generating happens while the GPUs are quiet, and that costs nothing at all.
Quiet, normal, busy
We rent the GPUs around the clock whether anyone is using them or not, so when they're idle we'd rather you had them.
Quiet
Hardly anyone is generating. Your generations are free and unlimited, and they don't touch your meter at all.
Normal
The usual state. A generation uses its model's standard amount of your allowance.
Busy
Everyone is generating at once. Generations use more of your allowance, or you can wait for things to go quiet and pay nothing.
The current state is always shown on the generate page, so you never have to guess.
Heavier models cost more
A small turbo model finishes in a second. A large one can take twenty times the GPU time for the same picture. If both cost the same, turbo users end up paying for everyone else, so usage scales with what the run actually costs.
| Model class | Examples | Relative cost |
|---|---|---|
| Turbo | Fast, low-step models | 1× |
| Standard | SDXL and similar | 3× |
| Heavy | Flux, large fine-tunes | 8× |
What each plan guarantees
A floor, not a ceiling. Quiet-hour generations are unlimited on every plan, including Free.
| Plan | Guaranteed generations / month | Models |
|---|---|---|
| Free | 1,000 | Turbo |
| Starter | 1,500 | Turbo and standard |
| Hobbyist | 3,000 | Everything |
| Pro | 10,000 | Everything, plus third-party APIs |
| Studio | 25,000 | Everything, including beta releases |
| Enterprise | Custom | Everything |
Discord server plans work the same way, except the pool is shared across everyone in the server.
If you run low
You'll watch your meter fill up long before it matters, and you have three ways out. None of them leave you stuck in the middle of a project:
- Wait for quiet hours. Free, unlimited, and always available.
- Top up. $5, $10, or $20 of extra usage added to your current cycle.
- Grab a pass or upgrade. A 1-day or 7-day Pro pass, or move up a plan.
Plans and prices open shortly. Discord hears first.
Where your money goes
Part of what you pay covers the GPUs. Another part goes to the person whose model you used: creators earn a royalty on every generation run with their model or LoRA. That's the whole point. People who publish good models should get paid when other people build on them.
Questions
It's real. We pay for the GPUs 24 hours a day whether they're working or not, so an idle GPU is money already spent. Letting you use it costs us nothing extra. The only catch is the obvious one: when everyone wants the GPUs at once, they aren't quiet any more.
No. Your allowance resets each billing cycle. Top-ups last until the end of the cycle you bought them in.
Paid generating pauses. Quiet-hour generating doesn't, so you can keep working for free whenever the GPUs are idle. Top up or upgrade if you'd rather not wait.
Because our costs do. When the queue is full, every generation competes for the same hardware. Charging more at the peak and nothing at the trough is what pays for a genuinely unlimited quiet period, instead of one hard cap on everyone all the time.
No. Pick a plan, generate, and watch the meter if you feel like it. This page exists because we'd rather explain how the machinery works than hide it.