Pay for what you use.
Nothing else.
No plans, no subscription, nothing that renews. Every new account starts with $15 of credit. AI is charged by the tokens each task actually uses, and apps on our servers by the memory, traffic and disk they actually use.
Free credit to start
Added when you sign up. No card, no trial clock. Enough for around ten typical tasks.
AI at cost
Tokens are charged at the model provider's own published rate. We don't add a markup on top.
Builds & deploys
Building your iOS and Android apps and deploying your web and API cost nothing. Only the AI work they trigger is billed.
AI models
USD per 1 million tokens. Pick the model in the console's model menu; you can switch per conversation.
| In the console | Model | Input | Cached input? | Output |
|---|---|---|---|---|
| Claude | Claude Opus 5 | $5.00 | $0.50 | $25.00 |
| DeepSeek | DeepSeek Flash | $0.30 | $0.006 | $1.20 |
| GLM | GLM-5.3 | $1.40 | $0.26 | $4.40 |
| Kimi | Kimi K3 | $3.00 | $0.30 | $15.00 |
| Gemini | Gemini 3 Pro | $2.00 | $0.20 | $12.00 |
| GPT | GPT-5.5 | $5.00 | $0.50 | $30.00 |
- Claude tasks are charged at the exact cost Claude Code reports when the task finishes, so a task that uses a different Claude model is priced as that model. The row above shows the default.
- DeepSeek has one rate around the clock: DeepSeek's peak-hour rate.
- Long-context surcharges (GPT-5.5 above 272K tokens, Gemini above 200K) are not passed on to you.
- Small housekeeping calls, like naming a conversation, summarizing a commit or sorting a build log, are free.
- Using your own API key? Those tasks run on your account with the provider. We charge nothing for them, and they don't count toward the daily limits below.
What does one task cost?
A task is one message you send, plus every step the AI takes to answer it: reading your code, writing files, running checks. It typically calls the model about ten times. Across 488 real Claude tasks we measured, the median cost was $1.29, and 9 in 10 cost under $6.32. A small fix costs cents; building a whole feature from scratch costs more.
You don't have to guess: the exact cost appears under every reply, and My Profile → Balance shows your balance and what you've spent today and this month.
App hosting
For apps deployed to our shared servers. Settled once a day (UTC), per app.
| Resource | Price | How it's measured |
|---|---|---|
| Memory | $0.005 / GB·hour | Your app's actual memory use, sampled every minute (about $3.60 per GB for a full month) |
| Outbound traffic | $0.10 / GB | Data your web pages and API send to visitors |
| Disk | $0.10 / GB·month | Your app's highest disk use each day, charged by the day |
- CPU is not billed. A day when an app uses nothing costs nothing and doesn't appear on your bill.
- For scale: a small app holding about 330 MB of memory around the clock comes to roughly $0.04 a day, about $1.20 a month.
- Apps on your own server (MyCloud) are not charged hosting by us. The machine is in your own cloud account and your provider bills you directly. How it works.
- These are starting rates. We'll review them against real usage, and any change will show up on this page.
How your balance works
Where credit comes from
$15 when you sign up, and $15 for each friend you invite once their first build or deploy completes (up to 3 friends). Online top-up isn't open yet. If you're running low, talk to us.
Starting and finishing a task
Starting a task, build or deploy needs at least $0.50 in your balance. A task that has already started always finishes. If it costs more than you had left, your balance dips below zero and the next task waits until it's topped up.
Daily limits (early access)
10 AI tasks, 6 builds, 6 deploys and 6 new apps a day, with 6 apps in total. Limits reset at 00:00 UTC. Tasks run on your own API key don't count. Need more room? Ask us.
Nothing that renews
There's no plan to cancel and nothing charged in the background. What you see under each reply, and in your Balance, is what you paid.
Building something bigger?
Higher limits, volume rates, or a server managed for you: contact us and we'll work out what fits your actual usage.
Contact Us →