Pay when cores work.
Strictly $0/hr when they sleep.
Stop paying always-on cloud tax for AI agents waiting on LLM tokens, idle preview apps, and quiet CI queues. Metered per active second with zero lock-in.
Metered strictly while your workload is executing instructions.
Zero charge for RAM while your VM is paused or snapshotted.
Pay only for unique changed blocks across database and sandbox forks.
Instant sub-second resume when an HTTP request, exec call, or user arrives.
Model your real workload bill.
Select a workload profile and dial in your vCPU, RAM, NVMe storage, and fleet size to compare active-usage billing against always-on cloud/PaaS rates.
Agents spend ~85% of wall-clock time waiting on LLM token generation. uncage.sh pauses compute to $0/hr between tool calls.
Comparison baseline: always-on 730h/mo cloud compute ($0.052/vCPU-hr, $0.011/GB-hr RAM, $0.15/GB-mo SSD). uncage.sh rates: $0.036/active vCPU-hr, $0.0054/active GB-hr, $0.10/GB-mo NVMe.
Predictable rates from 1 vCPU to 16 vCPU.
| Compute Spec | Included NVMe | Active Rate | AI Agent (15% Active) | 24/7 Active (730h) | Idle / Sleep | Launch |
|---|---|---|---|---|---|---|
| 1 vCPU · 1 GB | 10 GB | $0.0414/hr | $5.53/mo | $31.22/mo | $0.0000/hr | Deploy → |
| 2 vCPU · 4 GB | 20 GB | $0.0936/hr | $12.25/mo | $70.33/mo | $0.0000/hr | Deploy → |
| 4 vCPU · 8 GB | 40 GB | $0.1872/hr | $24.50/mo | $140.66/mo | $0.0000/hr | Deploy → |
| 8 vCPU · 16 GB | 80 GB | $0.3744/hr | $49.00/mo | $281.31/mo | $0.0000/hr | Deploy → |
Economy, Balanced, and Dedicated classes.
Match isolation and CPU scheduling to each workload without rewriting a single config line. Switch classes per service at any time.
| Capability | Economy Dev, Previews & Batch | Balanced
DEFAULT
Production Apps & Agents | Dedicated High-IOPS & Strict Isolation |
|---|---|---|---|
| CPU Allocation | Burstable high-frequency vCPU | Priority vCPU with burst headroom | Dedicated physical cores (zero contention) |
| Active Rate Multiplier | 1.0× base rate | 1.0× base rate | 1.35× base rate |
| Idle / Sleeping Compute | $0.00 / hr | $0.00 / hr | $0.00 / hr (when paused) |
| Boot & Memory Fork Speed | < 450ms resume | < 200ms boot & ~180ms live fork | < 200ms boot & pinned NVMe I/O |
| Regions Available | us-east, eu-central, us-west, ap-south | us-east, eu-central, us-west, ap-south | us-east, eu-central, us-west, ap-south |
| Database Backups & Branching | Daily snapshots + instant branching | Continuous PITR + instant branching | Continuous PITR + HA replica ready |
| Ideal Workloads | Dev sandboxes, PR previews, async jobs | Production APIs, AI agents, Postgres, CI | High-IOPS DBs, strict isolation fleets |
Billing & metering questions, answered.
01 How does per-second active billing work?
02 What happens to my bill when a sandbox, app, or preview database sleeps?
03 Are all five surfaces (Apps, Databases, Sandboxes, Desktops, CI) billed the same way?
04 How does the $10 starter credit work?
05 Can I set hard monthly spend limits so runaway AI agents never blow my budget?
Stop paying for sleeping containers and idle sandboxes.
Claim $10 in free starter credit and launch your first app, PostgreSQL database, or AI sandbox in under 30 seconds.