Update:Our Privacy Policy has changed

Pricing

Pay only for what you use with per-request pricing.
Or choose a Fixed plan for consistent usage with a flat price.

Free

$0

-

Perfect for prototypes and hobby projects.

Concurrent Boxes

10

CPU Hours / Month

5

$0.10

per active CPU hour

Pay for active time only. Boxes pause when idle.

Resources

2 vCPU, 4 GB RAM

Included Storage

10 GB

$0.0111

per box / hour

($8 / month)

Always-on boxes. One rate per hour, no usage charges.

Resources

2 vCPU, 4 GB RAM

Idle Timeout

Never

Compare Plans

Plans that scale to all sizes.

Free
-
Capacity
10
CPU Cores / Box2
Memory / Box4GB
Max Storage / Box10GB
CPU Hours / Month5
1 hour
RuntimesNode.js, Python, Go, Ruby, Rust
Price
Fixed PriceFree
Free
Memory PriceFree
Storage PriceFree
Start Now

FAQ

You are billed based on how much CPU your code actually consumes, measured in core-hours. For example, using 100% of 2 cores for one hour costs $0.2, while using 10% of a single core for one hour costs $0.01. When a standard box is idle with no CPU usage, no charges apply. Keep-alive boxes are billed at a fixed rate per hour by size instead.

  • Standard boxes automatically freeze after inactivity. Storage is preserved, but running processes are stopped until the next request.
  • When a new request comes in, the box wakes up automatically, similar to how AWS Lambda cold-starts on demand. No manual intervention is needed.
  • Keep-alive boxes stay on continuously and are billed at a fixed rate per hour by size instead of auto-pause behavior.

Keep Alive boxes are billed for the time they exist, at a fixed rate by size, billed per second. This is the only charge for that Keep Alive box, so you are not billed separately for CPU or storage on it.

  • Small Keep Alive box: $0.0111/hour ($8 per month)
  • Medium Keep Alive box: $0.0222/hour ($16 per month)
  • Large Keep Alive box: $0.0444/hour ($32 per month)

Storage is billed at $0.1 per GB per month and includes all data on disk, including snapshots. Storage limits are 10 GB for Small, 20 GB for Medium, and 30 GB for Large.

  • Up to 10 concurrent boxes.
  • 5 CPU hours per month.

  • Agents use your own model provider key (Anthropic, OpenAI, OpenRouter, ...). Set it on the box or save it on your account. Upstash does not provide or bill model usage.
  • Agent runs, scheduled runs and browser AI calls need a key. Shell, files, git, SSH, previews and browser CDP work without one.

Yes. The default quota for Pay as you go is 1,000 concurrent boxes. This is a soft limit that can be increased on request, similar to AWS service quotas. Just contact us and we'll raise your limit.

Boxes currently run on AWS us-east-1. We are planning to add more regions soon.

Yes, custom Docker container support is planned and coming soon.

Upstash Box pricing — all plans

Plan overview
PlanPriceConcurrent boxesCPU hours/month
Free$0105
Pay as You Go$0.10–$0.40 per active CPU hour1,000Unlimited
Fixed$0.0111–$0.0444 per box / hour1,000Unlimited
Box sizes (Pay as You Go) — all sizes
SizevCPUMemoryDiskUsage priceKeep-alive price
Small2 vCPU4 GB RAM10 GB$0.1/active CPU hour$0.0111/hour ($8/month)
Medium4 vCPU8 GB RAM30 GB$0.2/active CPU hour$0.0222/hour ($16/month)
Large8 vCPU16 GB RAM30 GB$0.4/active CPU hour$0.0444/hour ($32/month)

Pay as You Go — pricing details

  • Price: $0.10–$0.40 per active CPU hour
  • Concurrent boxes: 1,000
  • CPU hours/month: Unlimited
  • Storage: $0.10 per GB/month
  • CPU hour pricing: Small: $0.10, Medium: $0.20, Large: $0.40 per active CPU hour
  • Keep-alive pricing: Available per box at a fixed rate per hour. See the Fixed plan.

Fixed (Keep Alive) — pricing details

  • Price: $0.0111–$0.0444 per box / hour
  • By size: Small: $0.0111, Medium: $0.0222, Large: $0.0444 per box / hour ($8, $16, $32 per month)
  • CPU: Included in the fixed rate
  • Storage: Included
  • Idle timeout: none, boxes stay on continuously

Full machine-readable pricing is available at /pricing/box.md.