Two ways to pay. One rule: you pay for what you decide to do.
Darting picks models for you. When it does, that's on the plan. When you pick — a specific model, a video, an image — that's on your credits, at the provider's rate. Nothing else, no seats, no per-user fees.
$200/month
Everything Darting decides on its own, covered.
- Every Auto-routed turn, including the Intelligence tier
- Subagents spawned by tier — research, coding, review
- Sandbox time, web search, web fetch
- A monthly credit allowance for trying named models, video and image generation
Usage is paced in 5-hour windows and bounded weekly — the same shape you know from other AI subscriptions, so a heavy afternoon doesn’t lock you out for the month. When a window fills, you see exactly when it resets.
Need more? Stack it.A second plan doubles every budget on the same account — one click, one subscription, prorated to your billing date. Three plans triple it. There’s no “team tier” to upgrade into; the plan is the unit.
Or keep going on credits. Flip one switch and an exhausted window falls through to your credit balance instead of waiting for the reset. Your call, per account, any time.
Pay as you go
No subscription. Buy credits, use them at the provider’s API rate for everything Darting does. Credits are prepaid and never expire.
Credits are also what a named-model turn, a generated video, or a generated image bills to — with or without a plan. If you want to run every turn on a specific model, this is your track.
Side by side.
| Credits | Auto Plan | |
|---|---|---|
| Auto-routed work | Metered from credits | Included |
| Tier-spawned subagents | Metered from credits | Included |
| Sandbox, search, fetch | Metered from credits | Included |
| Named model picks | Credits | Credits (plan includes a monthly allowance) |
| Video / image generation | Credits | Credits (plan includes a monthly allowance) |
| Pacing | Your balance | 5-hour + weekly windows, stackable |
| Commitment | None | Monthly, cancel any time |
Questions people actually ask
What does “Auto” mean?
Darting routes each turn to the cheapest model that can do the job well, and escalates when the task needs it. You never choose; you never pay per token. If you do choose a model, that turn bills your credits instead — the plan covers Darting's decisions, not yours.
What happens when I hit a window?
The app pauses and shows the reset time. You can wait, add another plan, or switch on “continue on credits.” Nothing is lost; queued work resumes.
Do the windows reset at midnight?
No. The 5-hour window starts on your first use and resets five hours later. The weekly window is anchored to your first week on the plan, so resets spread across users instead of everyone hitting the same wall on Monday morning.
Can I go from two plans back to one?
Not mid-cycle — cancel at period end and resubscribe. Adding is one click; removing is one cancellation.
What about my team?
Every member has their own plan or credits. A project can bill its owner when a member runs out — so one person's exhausted window doesn't stall the team's work.
I already pay for Claude Max. Can Darting use that?
No — and you come out ahead. Darting runs a dozen agents at once in cloud sandboxes, which no personal subscription is built for. Instead you get every frontier model, Auto routing each step to the one that fits, and the machinery a terminal can't give you: a sandbox and branch per task, verification in a real browser, agents that fix their own CI and hand you a merged PR.
What you get insteadIs my code private, and where do agents run?
Yes. Each thread runs in its own isolated cloud sandbox, branched from your repo, and torn down after use. Private repos authenticate with a token you control. Heavy builds and test suites never touch your laptop.