Key facts
Inquir Compute Platform Specification
The main product facts in one place: what runs on the platform, with which defaults and limits. Other pages link here for the numbers; the documentation has the details.
Last updated: 2026-10-11
What Inquir Compute is
Inquir Compute is a cloud platform for applications, APIs, background workers and databases. Container applications are built from your Dockerfile on the platform or run from a ready image, and get health-checked releases with previews, promotion and rollback, a private network between services, persistent volumes, replicas, logs, resource metrics and alerts. Deploying straight from a GitHub repository is in early access, on request. Functions, API routes and pipelines with schedules are available in the same workspace.
Container applications: defaults and limits
Platform defaults per application and per workspace. Memory, CPU and live-release ceilings can be raised on request; your plan also sets a compute pool shared by all the workspaces you own.
| Area | Current default | Notes |
|---|---|---|
| Memory per release | 512 MB | Up to 2,048 MB; reserved in full and a hard limit. |
| CPU per release | 0.5 vCPU | Up to 2 vCPU; about half is reserved, the rest is burst headroom. |
| Replicas per application | 1 | Up to 8 in always mode; http and manual modes and applications with a writable volume run one. |
| Live releases per workspace | 10 | Previews, production and the rollback reserve together. |
| Startup health-check deadline | 120 s | A release that does not pass in time never receives traffic. |
| Window after a release change | 5 min | The previous production release keeps running as an instant-rollback reserve. |
| Preview release lifetime | 24 h | Releases that were not promoted are stopped after this. |
| Public TCP ports | Not available at the moment | On the private network a service keeps its container port. |
Networking, databases and volumes
- Every application has a private address, {app}-{suffix}.apps.internal, reachable only from the other applications of the workspace; it survives redeploys and rollbacks.
- Public access is opt-in: a new application has no public address until you generate one. Custom domains are verified with DNS records and get TLS automatically.
- Database templates: Postgres 16 and Redis 7, each with a persistent volume, a private TCP endpoint and a generated password. Connect one to a service and DATABASE_URL or REDIS_URL lands in its variables.
- A persistent volume lives on one server: an application with a writable volume runs one replica and is recreated on deploy, with a few seconds of downtime. Volumes are backed up automatically on a schedule, every 6 hours today (databases as crash-consistent snapshots with a brief freeze); applications whose volumes hold more than 2 GB in total are not included yet.
Function runtimes
| Runtime | Version |
|---|---|
| Node.js | 22 |
| Python | 3.12 |
| Go | 1.22 |
Function execution model and limits
| Area | Current default | Notes |
|---|---|---|
| Function timeout | 5,000 ms | Per-function setting, up to 24 hours (86,400,000 ms). |
| Function memory | 256 MB | Per-function setting in the editor. |
| Isolation | Container per function | Each function gets its own containers from a shared runtime image, with its code mounted read-only. |
| Warm containers | Supported | Repeat calls reuse a running container; the first call after a quiet period can still start cold. |
Gateway model
- One API gateway with route-based dispatch to functions.
- Route config includes method, path pattern, auth mode, and related options.
- Supports use cases like REST APIs and webhook handlers.
Pipelines and background work
- Pipelines can model scheduled jobs, async workflows, and multi-step automation.
- Graph editor supports branching, mapping, and human-in-the-loop steps.
- Runs are visible in execution history with logs and traces.
Observability and security defaults
- Execution history includes status, timing, and logs.
- Trace and span data is available for supported runtimes.
- Per-function environment variables and secret-like config values are supported.