Console: 2 cores, 4 GB, 40 GB. Host: 8 cores, 16 GB, 100 GB.
FREE CAPACITY PLANNING TOOL
MonkeyCode self-hosting capacity calculator.
Estimate a planning envelope for a pilot. Published minimums and editorial assumptions are labeled separately so platform teams can challenge every input.
PLANNING BASELINE1 development host
- Total CPU
- 10 cores
- Total memory
- 20 GB
- Total storage
- 200 GB
Includes the published console minimum and editorial host assumptions.
Transparent formula
A planning assumption, not a sizing promise.
Assumptions version 2026-07-15. Replace them with measured pilot telemetry before production.
| Host count | ceil(concurrency ÷ 4) |
|---|---|
| Host CPU | max(hosts × 8, concurrency × profile CPU) |
| Host memory | max(hosts × 16, concurrency × profile memory) |
| Host storage | max(hosts × 100, concurrency × retained GB) |
| Total | Host envelope plus the 2-core / 4 GB / 40 GB console floor |
Four environments per host; light 1C/2 GB, standard 2C/4 GB, heavy 4C/8 GB per concurrent environment.
No throughput, latency, build duration, availability, or isolation guarantee is implied.
Static examples
Standard profile with 40 GB retained per environment.
These examples use the same shared calculation as the interactive result.
| Concurrency | Hosts | Total CPU | Total memory | Total storage |
|---|---|---|---|---|
| 1 | 1 | 10 cores | 20 GB | 140 GB |
| 4 | 1 | 10 cores | 20 GB | 200 GB |
| 10 | 3 | 26 cores | 52 GB | 440 GB |
Official minimum source: MonkeyCode README ↗
How does the capacity calculator estimate resources?
It sizes hosts as ceil(concurrency divided by 4), then takes the larger of a per-host floor (8 cores / 16 GB / 100 GB) or per-environment demand for the chosen workload profile, and adds the published 2-core / 4 GB / 40 GB console minimum. Published floors and editorial assumptions are labeled separately so you can change them.
Is the result a guaranteed production sizing?
No. It is a transparent planning baseline for a pilot, not a throughput, latency, availability, or isolation guarantee. It excludes stopped environments, model inference, high availability, backups, log and image growth, and burst headroom—replace the assumptions with measured pilot telemetry before production.
Where do the minimum figures come from?
The console and development-host minimums come from the public MonkeyCode README; the per-host and per-environment sizing factors are clearly labeled editorial estimates you can challenge and adjust.