| Self-host difficulty | EasyFIRST-PARTY: Node 22 + npm i -g @openai/codex + OpenAI API key (the exact steps our preset runs) |
|---|
| Deploy | Native, BinaryRust binary; npm/brew install; confirm first-party |
|---|
| Scaling | Single process |
|---|
| Config complexity | TrivialFIRST-PARTY: measured on a DigitalOcean s-1vcpu-1gb (1 vCPU / 1 GB, Ubuntu 22.04), harness run 20260826-195645, 2026-08-26. Node 22 plus one global npm package, no config file and no service unit; configuration is an API key. Installed within our 60s polling resolution. |
|---|
| Min vCPU (measured) | 1 vCPUFIRST-PARTY: measured on a DigitalOcean s-1vcpu-1gb (1 vCPU / 1 GB, Ubuntu 22.04), harness run 20260826-195645, 2026-08-26. Installed and verified on a single vCPU. |
|---|
| Min RAM (measured) | 1 GBFIRST-PARTY: measured on a DigitalOcean s-1vcpu-1gb (1 vCPU / 1 GB, Ubuntu 22.04), harness run 20260826-195645, 2026-08-26. Proven on a 1 GB box, with the same caveat as claude-code: the harness installs and verifies codex but never drives a session, so this is the floor for installed-and-ready, not for a running task. |
|---|
| Idle RAM (measured) | None residentFIRST-PARTY: measured on a DigitalOcean s-1vcpu-1gb (1 vCPU / 1 GB, Ubuntu 22.04), harness run 20260826-195645, 2026-08-26. Nothing resident. Like claude-code, codex is invoked per session rather than run as a service; the idle box measured 323 MB, essentially all OS, our agent (~68 MB) and Caddy (~50 MB). |
|---|
| Infra cost | Very lowFIRST-PARTY: measured on a DigitalOcean s-1vcpu-1gb (1 vCPU / 1 GB, Ubuntu 22.04), harness run 20260826-195645, 2026-08-26. No resident process, so no standing cost beyond the box. |
|---|
| One-click on AgentOcean | YesFIRST-PARTY: production install preset in src/lib/products.ts (id: codex); BYO OpenAI key |
|---|