Stack calculator
Pick the apps you want on one server. We add up their documented RAM figures and show the cheapest plans that clear the total.
Where the numbers come from
Until our own measurements are published, the calculator uses each project's documented figure for a whole server. That makes the total conservative: each figure already includes room for the operating system, so a stack of three apps rarely needs the full sum. Treat the result as a safe upper bound, not a tight fit.
| App option | RAM used | Min vCPU | Basis |
|---|---|---|---|
| Immich (machine learning on) | 6 GB | 2 | Official minimum (6 GB, 2 cores). Source |
| Immich (machine learning off) | 4 GB | 2 | Official: runs on 4 GB with ML disabled. Source |
| Nextcloud AIO (no heavy add-ons) | 3 GB + 1 GB per active user | 2 + 1 per active user | AIO minimum 2 GB + recommended 1 GB headroom, plus 1 GB and 1 vCore per active user. Source |
| Nextcloud AIO + ClamAV / full-text search | 4 GB + 1 GB per active user | 2 + 1 per active user | AIO minimum 3 GB with ClamAV/Fulltextsearch + recommended 1 GB, plus 1 GB and 1 vCore per active user. Source |
| n8n (core, no Assistant sandbox) | 2 GB | 1 | Top of n8n's documented 320 MB – 2 GB memory example. Source |
| n8n (official compose with Assistant sandbox) | 4 GB | 2 | Official Docker Compose minimum (4 GB, 2 vCPUs). Source |
| Coolify (official minimum) | 2 GB | 2 | Official minimum (2 cores, 2 GB RAM). Source |
| Coolify, building apps on the same server | 4 GB | 2 | VPSfit recommendation: 2 GB servers stalled building ordinary apps (GitHub #2088), and Coolify's own 4 GB example server averages 2.6 GB with builds done elsewhere. Source |
vCPU: we use the largest single-app minimum (plus per-user cores), not the sum, because idle apps share cores well. Prices checked Oct 21, 2026.