Pricing

Start free. Pay for what runs.

One subscription per organization, plus metered compute and storage. Upgrade or cancel at any time.

View plans
2 months free on yearly
Free
For trying things out.
$0/month
  • 3 projects
  • 137 vCPU-hours of compute per month
  • Apps sleep when idle
  • Always-on with a card on file
  • 500 build minutes per month
  • 100 GB bandwidth per month
  • 3 working AI fixes per month
  • Up to 10 AI fix attempts per month
  • Custom domains
  • 5-day database restore points
  • Unlimited team members
Pro
Popular
For growing teams.
$19/month
14-day free trial
  • Unlimited projects
  • Always-on apps, pay-as-you-go compute
  • 20 AI build fixes per month
  • 500K AI assistant tokens per month
  • 3000 build minutes per month
  • 500 GB bandwidth per month
  • Custom domains
  • 30-day database restore points
  • Unlimited team members
Enterprise
For organizations at scale.
$99/month
  • Unlimited projects
  • Always-on apps, pay-as-you-go compute
  • 100 AI build fixes per month
  • 2M AI assistant tokens per month
  • 10000 build minutes per month
  • 2000 GB bandwidth per month
  • Custom domains
  • Unlimited team members
  • 60-day database restore points
  • Priority support
Usage-based resources

Metered, not guessed at.

Each plan includes a monthly allowance of build minutes and bandwidth, shown on its card above. AI auto-fix — Free: 3 working AI fixes per month, up to 10 attempts per month; Pro: 20 working AI fixes per month, up to 100 attempts per month; Enterprise: 100 working AI fixes per month, up to 500 attempts per month — you only pay for auto-fix when it works, and a failed attempt is never charged, but every attempt, working or not, counts toward the attempt limit. Every build that runs counts toward the build minutes — failed, canceled and replaced builds included — except an auto-fix build whose fix does not work, and a build that failed because of the platform itself; neither is ever charged. With a card on file, build minutes and bandwidth past the allowance are billed per unit; without one they stop at it until the next month or a card is added. Always-on compute is billed pay-as-you-go — apps that are not always-on sleep when idle instead.

Compute$0.04 per vCPU-hour