Your people are free. Pay for what the app runs.
One public rate card covers compute, data, network, observability, and agent operations. No plans or seat licenses hide the real bill. Your invoice is each meter below times its measured usage.
$5/mominimum, credited against usage
| Meter | Rate |
|---|---|
| Compute | |
| Ephemeral execution shared CPU | $0.000004 / vCPU-second |
| Ephemeral execution performance CPU | $0.00006 / vCPU-second |
| Ephemeral execution memory | $0.000009 / GB-second |
| Always-on compute (shared, 1 vCPU / 2 GB) | $0.042 / hour |
| Always-on compute (shared, 1 vCPU / 1 GB) | $0.03 / hour |
| Always-on compute (dedicated, 1 vCPU / 2 GB) | $0.125 / hour |
| Scale-to-zero requests | $0.9 / million requests |
| Network & object storage | |
| Egress | $0.06 / GB |
| Object storage | $0.045 / GB-month |
| Databases & recovery | |
| Database storage overage | $0.3 / GB-month |
| Extended backup storage | $0.045 / GB-month |
| Managed database db-n1, single node (1 shared vCPU / 512 MB RAM, 10 GB disk) | $5 / month |
| Managed database db-n1, high availability (1 shared vCPU / 512 MB RAM, 10 GB disk) | $49 / month |
| Managed database db-s1, single node (1 shared vCPU / 1 GB RAM, 10 GB disk) | $19 / month |
| Managed database db-s1, high availability (1 shared vCPU / 1 GB RAM, 10 GB disk) | $79 / month |
| Managed database db-s2, single node (2 shared vCPU / 4 GB RAM, 20 GB disk) | $72 / month |
| Managed database db-s2, high availability (2 shared vCPU / 4 GB RAM, 20 GB disk) | $229 / month |
| Managed database db-s4, single node (4 shared vCPU / 8 GB RAM, 40 GB disk) | $144 / month |
| Managed database db-s4, high availability (4 shared vCPU / 8 GB RAM, 40 GB disk) | $449 / month |
| Managed database db-p1, single node (1 performance vCPU / 8 GB RAM, 80 GB disk) | $119 / month |
| Managed database db-p1, high availability (1 performance vCPU / 8 GB RAM, 80 GB disk) | $369 / month |
| Managed database db-p2, single node (2 performance vCPU / 16 GB RAM, 160 GB disk) | $219 / month |
| Managed database db-p2, high availability (2 performance vCPU / 16 GB RAM, 160 GB disk) | $709 / month |
| Managed caches | |
| Managed cache cache-n1 (256 MB memory) | $6 / month |
| Managed cache cache-s1 (1 GB memory) | $18 / month |
| Managed cache cache-s2 (4 GB memory) | $64 / month |
| Managed realtime | |
| Realtime messages (published plus delivered)First 1 1M messages included each organization-month | $2.5 / 1M messages |
| Realtime connection timeFirst 1 1M minutes included each organization-month | $1 / 1M minutes |
| Realtime retained payload | $0.05 / GB-day |
| Queues & scheduling | |
| Queue operations (64 KB chunks)First 1 1M operations included each organization-month | $1.2 / 1M operations |
| Queue retained payloadFirst 30 GB-day included each organization-month | $0.0015 / GB-day |
| Cron executionsFirst 100 1K executions included each organization-month | $0.1 / 1K executions |
| Observability & assistance | |
| Extended metrics retention | $0.2 / GB-month |
| Speed Insights data pointsFirst 1 100K data points included each organization-month | $0.3 / 100K data points |
| Log events | $1.7 / 1M events |
| Platform agent run | $0.2 / run |
| AI gateway | |
| AI gateway inference | $1.15 / $1 direct model usage |
Metered continuously, billed monthly. The $5 minimum is credited against everything above.
Worked examples
What real workloads cost
Line items straight from the rate card — totals computed, never quoted.
Hobby full-stack app
Scale-to-zero app with a small managed database and light traffic.
Scale-to-zero requests2 × $0.9 / million requests
Egress5 × $0.06 / GB
Object storage2 × $0.045 / GB-month
Managed database db-s1, single node (1 shared vCPU / 1 GB RAM, 10 GB disk)1 × $19 / month
Log events2 × $1.7 / 1M events
Estimated monthly$24.59/mo
Production API
Two always-on instances with a 10 GB managed database.
Always-on compute (shared, 1 vCPU / 2 GB)1,440 × $0.042 / hour
Egress50 × $0.06 / GB
Managed database db-s2, single node (2 shared vCPU / 4 GB RAM, 20 GB disk)1 × $72 / month
Log events5 × $1.7 / 1M events
Estimated monthly$143.98/mo
FAQ
Questions, answered
Free tiers are paid for by someone — usually with throttled performance, surprise limits, and upsell walls. Korve charges honestly instead: a $5/mo minimum that is fully credited against your usage, so a small project effectively runs on the minimum alone.
Every resource maps to a public meter: compute by the hour, scale-to-zero and queue operations by the million requests, cron executions by the thousand, egress and overage storage by the GB, and logs by event count. Meters accrue continuously and your invoice is the rate card times what you ran — to the cent.
Each database SKU includes the disk shown in its name; database storage is billed only above that amount. Every database also includes 14 days of point-in-time recovery, 30 daily backups, 12 weekly backups, and 12 monthly backups. Longer managed retention uses the extended-backup meter. Customer-owned backup destinations are not charged managed storage; outbound transfer uses the public egress meter.
Yes. Set a monthly budget per project and choose alert-only, which is the default, or a hard stop. With the hard stop on, Korve refuses new deploys — including the ones a Git push triggers — once month-to-date spend reaches the cap, while already-running workloads keep serving until you change or remove them.
Same meters, same rates — whether a deploy comes from the dashboard, the CLI, or an agent over MCP. The only agent-specific meter is platform diagnosis, billed at $0.2 / run when you ask it to investigate a failure.